Finding Potential Speed Bumps and Pitfalls in Buyer–Seller Negotiations in Twenty Cultures

With the advent of money came a distinction between buyers and sellers, a distinction loaded with cultural nuance. In American commerce we have the dialectic: “let the buyer beware” versus “the customer is always right.” Indeed, both sound correct. But the Japanese corollary is different: “Everywhere the buyer is king, but in Japan the buyer is ‘kinger.’” That is, hierarchy is paramount in Japanese society, and buyer–seller relationships are characterized more by indulgent dependency (amae) than in all other countries. Typically, Japanese sellers make long presentations, then accede to Japanese buyers’ wishes without objections because they can trust buyers to take care of them in the long run.

This article takes a broader view of such cultural differences and their impact on buyer–seller negotiation processes and outcomes. Such differences often manifest themselves as “speed bumps” or even “pitfalls” in international commercial negotiations.

Following a similar path as our previous work (cf. Graham, Mintu‐Wimsat, and Rodgers 1994), the fundamental theory underpinning our study suggests that culture affects negotiation behaviors and outcomes and it moderates the relationships (i.e., strengthens or weakens or even reverses the valences) between such behaviors and outcomes. See Figure One. Most research in the area concerns the effects of culture on negotiation behaviors and outcomes but not the relationships between the two.

The study also integrates theories and methods from a range of behavioral sciences: marketing science, decision analysis, behavioral economics, game theory, social psychology, anthropology, sociolinguistics, linguistics, content analysis, and structural equations modeling. The data were created in laboratory settings in which 1,197 experienced businesspeople from 20 cultural groups participated in a three‐product buyer–seller negotiation simulation. Identical methods were used in each of the data collection locations—most of which were outside the United States—and the data were compiled over a span of 36 years. (For details, see the section on limitations and future research below and the  Appendix).

This article begins with introductory and background information, followed by a description of how our database were developed. We then take a quick look at how observed behaviors are associated with questionnaire‐derived measures of negotiation processes and outcomes. This is followed by an investigation of how six previously identified indices of culture predict negotiation behaviors, processes, and outcomes across the twenty cultural groups included in our database. We also describe and utilize a new construct, a “latent” one, that encapsulates a variety of seemingly disparate cultural indices. We then discuss limitations and suggestions for future research and end the article with our conclusions.

It is important to acknowledge at the outset the risk that this type of work may foster and perpetuate cultural stereotypes. That is certainly not our intention. Although it is useful to identify important differences across cultural groups, no cultural group is completely homogeneous in any of the behaviors and outcomes measured in our study. Little replication is done in this field and readers should be cautious in interpreting and applying our data. Our findings are most useful as guides for future research and as signposts of potential difficulties in cross‐cultural negotiations.

The interdisciplinary nature of international negotiation yields a hodgepodge of disparate contributions. To provide a context for our broad empirical studies of international buyer‐seller negotiations, we briefly summarize and comment on the relevant literature, which has both great value and substantial weaknesses.

Despite the importance of the topic (cf. Salacuse 2010), there are few empirical studies of the effects of culture on negotiation (e.g., Reynolds, Simintiras, and Vlachou 2003; Lewicki, Barry, and Saunders 2016). Moreover, most of the “scientific” literature is narrowly focused with respect to constructs, countries/cultures considered, and analytical techniques (usually ANOVA). Experimental designs are generally ill‐fitted for comprehensive examination of the phenomenon. Until quite recently little attention has been paid to culture as a moderating effect. Moreover, the foci are almost always on transactions and economic outcomes, not long‐term relationships, trust as an outcome, and customer loyalty. Longitudinal studies are rare. Frequently, convenience has driven sample selection. Indeed, much of our own previous work is subject to these criticisms.

For example, prominent researchers have reported that negotiators from more collectivistic cultures have been found to achieve higher joint outcomes (Lituchy 1997; Arunachalam, Wall, and Chan 1998). Relatedly, Liu and Wilson found an important relationship between integrative issue‐linking (a kind of holistic approach) and higher joint gains. Complex patterns of difference in negotiation outcomes across cultures have been reported by Brett et al. (1998) and Brett (2001). Generally, cross‐cultural negotiations produce lower joint economic outcomes than intracultural ones (e.g., Adler and Graham 1989; Natlandsmyr and Rognes 1995; Brett and Okumura 1998).

Lewicki, Barry, and Saunders (2016) found that culture has been shown to influence a variety of negotiation processes, including: concession making (Faure 1999); the use of representational strategies (Adler, Graham, and Schwarz 1987); information exchange (Adair, Okumura, and Brett 2001); interruptions (Adler, Brahm, and Graham 1992); and extreme offers (Gelfand and Christakopoulou 1999). The work of both Hall (1976) and Hofstede (2003) has proved useful in predicting the observed behavioral differences in several of the studies in this area (e.g., Reynolds, Simintiras, and Vlachou 2003; Samaha, Beck, and Palmatier 2014).

Aslani et al. (2016) reported that negotiators from different cultures tend to rely on different negotiation strategies with concomitant outcomes. For example, Middle Easterners and Chinese more frequently take a competitive approach and Americans more often a cooperative one. Liu and Wilson (2011) found the same distinction between American and Chinese negotiators. Others have reported that Americans are more competitive than Chinese (cf. Adler, Brahm, and Graham 1992).

Cultural differences in the perception and valuing of time were first elucidated by Hall (1960). Armagan et al. (2006) detailed how such differences in “temporality” affected negotiation behaviors and outcomes in Portugal, Turkey, and the United States. Salmon et al. (2016) showed that impatience among “individualistic” American participants (compared to Lebanese) reduced profits in intracultural laboratory experiments.

More recent studies have demonstrated that often‐used negotiation strategies do not always produce the expected results because of moderating effects, as represented in Figure One above. The tactics to outcomes relationship may be moderated by many factors, including whether the negotiation is intracultural or intercultural (Liu et al. 2012) and whether the participants have a holistic or analytical mindset (Brett, Gunia, and Teucher 2017). Gelfand et al. (2015) introduced the useful linguistic concept of “honor talk,” set forth ways to measure it, reported variation in the content of honor talk between Americans and Egyptians, and demonstrated its analytical value as a moderator effect.

We agree with those who criticize the empirical literature for tending to ignore the social context of international negotiations (Jönsson 2015). Zaltman (1997) voiced a similar complaint about marketing research in general. Indeed, the emphasis by those working in the area has been on rational processes (e.g., game theory) rather than on emotions (exceptions include Graham 1990; Gelfand and Brett 2004; Lee, Yang, and Graham 2006) and feelings of interpersonal (not physical) attraction and relationships.

We do see great value in the comprehensive approaches taken by Salacuse (2003) comparing 12 countries on ten cultural dimensions; Katz (2017) providing briefings on 50 countries; Meyer (2014) looking at around 25 countries on eight dimensions; and Usunier (2019) comparing 12 cultural groups on (in his words) “salient” dimensions. They mostly rely on their own observations, excellent scholarship (referring to the classics, e.g., Hall 1976 and Hofstede 2003), and self‐report data. However, incomplete descriptions of their methods are common in this genre. Our present study is more systematic, and, therefore, directly subject to replication and verification.

Finally, we are most pleased to see an increasing interest in notions of creativity in negotiations (e.g., Crotty and Brett 2012; Graham, Lawrence, and Requejo 2014; Gelfand et al. 2015; Aslani et al. 2016; Graham 2018). While many authors in the area describe the value of brainstorming, the broader topic of creativity, so crucial to long‐term buyer–seller international relationships, is generally ignored in the negotiation literature.

We collected data representing how businesspeople in 20 cultural groups behave in a buyer–seller negotiation. The database contains valuable information on the individual cultures. These data also provide a new way to discover patterns of variation across cultures in negotiation behaviors, processes, and outcomes.

A crucial framework for our study is “buyer‐seller negotiations”—also called “marketing negotiations”—although many of our research findings are generalizable to other social‐exchange settings such as labor–management disputes, political negotiations (e.g., arms or trade treaties), and even household divisions of labor. However, three characteristics of marketing negotiations limit the generalizability of our findings. First, in buyer–seller negotiations money (an easily quantifiable economic good) is exchanged for products or services. Second, the roles of buyers and sellers are often imbued with distinct responsibilities. For example, sellers almost always make first offers. Also, more risk is assumed by buyers because payments to sellers are “money in the bank” while costs of holding inventories tend to be more volatile. Third, the frameworks of conflict resolution, problem‐solving, integrative bargaining, and win–win negotiations that we have used in this research tend to be short‐term and transaction oriented. Such frameworks are generally not useful when analyzing international commerce, in which most often the embedded diversity leads to inventive, long‐term relationships and the potential for the fastest human progress (cf. Diamond 1997). As an example, consider the decades‐long history of the Apple‐Foxxconn relationship that produced the smart phone, the supply chain for which connects ideas and companies from Japan, Germany, South Korea, Taiwan, China, and the United States. The relationship led to successful inventions and increased prosperity (for some), all through an international network of buyer–seller negotiations and relationships.

A second fundamental framework for this study is a six‐construct structural equations model (SEM) that tracks processes and outcomes of marketing negotiations. The validity of this problem‐solving model of buyer–seller negotiations was explored and proven by Graham, Mintu‐Wimsat, and Rodgers (1994). The theoretical model employed is represented in Figure Two.

The key construct in the model is a problem‐solving approach (PSA) to negotiation. Among its several conceptually overlapping synonyms such as representational, cooperative, direct/open bargaining, question & answer (Q&A), and soft‐line strategies, currently integrative bargaining (or integrative negotiation) is perhaps the most commonly used label. The PSA approach emphasizes asking questions and exchanging information about negotiators’ needs and preferences in order to achieve mutually beneficial outcomes. The conceptual opposite to integrative bargaining is distributive bargaining, which is sometimes referred to as individualistic or competitive, substantiation and offers (S&O), or hard‐line approaches to negotiation (cf. Pruitt 1981). Both sets of terms and the structural model represent a transactional view wherein a negotiation is thought of as an event that is ideally concluded with an agreement.

Individual profits comprise the economic outcome. The model also includes relational constructs—interpersonal attraction and negotiator satisfaction, both of which hold implications beyond the immediate transaction. Support for the six hypothetical causal paths was described in great detail in Graham, Mintu‐Wimsat, and Rodgers (1994) and is summarized below:

Graham, Mintu‐Wimsat, and Rodgers (1994) tested the model using a marketing negotiation simulation—Kelley’s three‐product, mixed‐motive, buyer‐seller game (1966)—in which 700 businesspeople from 11 cultures participated. Subsequently, similar data were collected in nine additional countries and regions. The simulation included bargaining over the prices of lots of three products, each price level coinciding with different profit levels for buyers and sellers. Logrolling allowed for larger joint profits, but each bargainer in the dyad achieved separate, and usually different profits. There was a 1‐hour time limit and the participants spent an average of 35 minutes negotiating.

All participants were at least 25 years old and had at least 2 years of full‐time work experience in their respective countries (see the  Appendix for details). The average age across groups was over 30 years and the average work experience was over 7 years. All negotiators were participants in either MBA or executive education programs and volunteered for the research project.

Following the simulations, both buyers and sellers completed identical questionnaires yielding measures of the six constructs in the model (see Figure Two). Each person recorded the agreement reached (corresponding to individual profit levels achieved) and role played (buyer or seller). A negotiator’s problem‐solving approach (PSAn) was measured using three self‐report 5‐point Likert scales, the ends of which were for example, “solving a mutual problem” versus “self‐interested.” These three measures were combined with three from the partner’s questionnaire to compose a single case for analysis: the partner’s self‐reported PSAp; the partner’s self‐reported satisfaction (SATp), four 5‐point Likert scales (e.g., “satisfied” versus “dissatisfied”); and the negotiator’s attractiveness (ATTn), three items rated by the partner (e.g., “comfortable” versus “uncomfortable”). Thus, one buyer–seller dyad produced two cases for the analyses. See Graham, Mintu‐Wimsat, and Rodgers (1994) for a detailed description of the measurement methods used.

Both theoretical and measurement issues were addressed using structural equations modeling (SEM) and partial least squares (PLS) as the primary data analysis approach. We took this approach for four reasons. First, the phenomena are better represented by structural equations modeling wherein path coefficients are estimated simultaneously. Traditional path analysis is less rigorous because of the reductionism inherent in its sequential analysis approach. Second, in PLS parameters can be estimated independent of sample size. Given that sample sizes vary from 160 to 25, PLS seems the more appropriate method for parameter comparisons across cultural groups than LISREL. Third, PLS avoids parameter estimation biases inherent in regression analysis (Fornell, Rhee, and Yi 1991) and some of the restrictive assumptions underlying LISREL (Fornell and Bookstein 1982). Fourth, and perhaps most important, PLS provides the most flexibility regarding measurement of constructs. It offers a unique formative indicator measurement approach that can serve to mitigate the inherent imprecision in questionnaire translation processes. All four of these arguments are fully elaborated in Graham, Mintu‐Wimsat, and Rodgers (1994).

The results in the 1994 study regarding the universality of the model first developed in the United States proved equivocal; findings varied across the 11 cultural groups in most cases. However, the theoretical model still appeared to be a useful tool for understanding how business negotiations vary across cultural groups.

The data for the nine additional countries and regions were collected using identical methods. Analyses of the data from three cultures—Brazil (n = 70), Japan (n = 44), and Spain (n = 56)—were reported in Graham and Mintu‐Wimsat (1997). Graham (1995) reported results from a similar study in Norway (n = 32). Here we report new results from similar studies in southern China (Guangzhou area, n = 25), Hong Kong (n = 44), the Czech Republic (n = 46), the Philippines (n = 94), and Iran (n = 86). The data for all 20 cultural groups were collected over a 36‐year period. (See the  Appendix for the dates of data collection in each country.) Despite the long span of the data collection we see little evidence of history effects. We discuss this issue in detail in our section on limitations below. Thus, in Tables One and Two we display results based on a total of 1,197 participants from 20 different cultures and locations. The unit of analysis is the cultural group. In the  Appendix we list the demographics of the 20 cultural groups included in the study.

As shown in Tables One and Two, negotiation processes and outcomes varied widely across the 20 cultural groups. Indeed, a fundamental finding is the great diversity in how negotiation rituals work across the groups. However, one cultural universal is evident in the Path 5 data in Table Two.

In all 20 cases negotiators were found to be more satisfied with outcomes when they rated their partners as more interpersonally attractive. Others have pointed out similar universalities (e.g., Gunia et al. 2013). Certainly some of this consistency is a consequence of methods bias—the SATp and ATTn items were adjacent on the questionnaires. However, if we (rather arbitrarily) subtract .20 from each parameter estimate to compensate for potential methods bias, we still find 14 path coefficients greater than .20. So, despite this potential methodological weakness, Path 5 is very well supported.

In addition to highlighting the rich diversity in cultural patterns, other insights emerged from our work.

First, you may recall from our literature review that we found a discrepancy across studies. Alsani et al. (2016) and Liu and Wilson (2011) reported Chinese negotiators to behave more competitively in simulated negotiations than Americans, and we reported the opposite in Adler, Brahm, and Graham (1992). Perhaps the explanation has to do with history effects or regional differences in data collection. That is, our data were collected two decades before the other two studies. Meanwhile, in the two more recent studies, the region that the Chinese subjects came from was unspecified. Moreover, in the Liu and Wilson study, the Chinese subjects had been attending university in the United States for up to 5 years. The important point here is the necessity to account for the cultural diversity within China such as that evinced in Tables One and Two across the data we collected separately in Tianjin, Guangzhou, Hong Kong, and Taiwan. We also note that in Table One our American participants, mainly from the West Coast of the United States, were rated less cooperative (i.e., lower PSA) than any of the four groups from Greater China.

Second, both researchers and practitioners will be interested in comparing pairs of cultural groups. Large differences are indicants of “speed bumps and pitfalls” in associated intercultural negotiations. For example, comparing the data from the Iranian and American groups, similarities are evident, but more important, differences in negotiation rituals are illuminated. The Americans’ individual profits ($) are mediocre and the Iranians’ individual profits are among the very lowest. Skipping to the far right, the role of buyer versus seller proved to be important for the American group, but not the Iranian. In Tables Two, Path 2 partners’ satisfaction (SATp) was influenced by negotiators’ problem‐solving approach (PSAn) in the U.S. group, but not the Iranian one. (Welch’s t‐tests were conducted by conservatively estimating American standard deviations to be three times the Iranian ones for each path coefficient.) The data allows for comparisons of 190 pairs of cultural groups.

Finally, in Tables Two we can see how useful is the PSA model of negotiation portrayed in Figure Two by comparing the R² statistics for the two outcome measures—negotiators’ profits ($n) and partners’ satisfaction (SATp)—across the cultural groups. The variances in negotiators’ profits ($n) are best explained for the South Korean, Filipino, and Japanese groups. The model does poorly in explaining negotiators’ profits for those in Guangzhou PRC, Brazil, and Francophone Canada. The model works much better in explaining the variance in partners’ satisfaction (SATp) across the groups—Guangzhou PRC, Russia, and the UK are all well over 50 percent.

Using the approach detailed in Graham (1985 and 1993), we studied the verbal behaviors of negotiators in fifteen of the cultures. Six negotiators in each of the 15 groups were randomly selected from among larger groups as volunteers for videotaping. Employing a content analysis scheme developed by Angelmar and Stern (1978) for studying bargaining in marketing settings, 11 verbal behaviors were coded using transcripts of the videotaped negotiations. The numbers in Table Three are the percentages of statements that were classified into each category. That is, on average for the six Brazilian negotiators, 3 percent of their statements were classified as promises, 2 percent were threats, 22 percent were questions, and so on.

We recognize that six participants cannot possibly represent the cultural diversity of an entire region or country. Indeed, neither can 30 or even 160 participants. However, given the expenses of time and money in creating and analyzing videotape data, we think this is a reasonable start in going beyond the survey and experimental methods and measures typical in the negotiation research area. Also, in another study (cf. Roemer, Neu, Garb, and Graham 1999) we conducted similar analyses of larger numbers of American (n = 30) and Russian (n = 26) sellers in the simulation and compared them to our analyses reported here wherein n = 6 for both groups. The correlations between the larger and smaller samples are greater than r = .984, p < 0.001 for the arrays of behaviors for both the Russian and American groups.

We also used the transcripts and direct observations of the videotapes to develop six sociolinguistic measures of the interactions: the frequencies of the words “no” and “you,” silent periods of 10 or more seconds, conversational overlaps (instances of simultaneous talking), and touching. Also, the time of facial gazing (per 30 y minutes of interaction) was coded for each negotiator. All the numbers provided in Table Four represent averages for the six negotiators in each of the cultural groups. While our work here merely scratches the surface of these kinds of behavioral analyses, it does indicate substantial cultural differences.

While Table Three clearly shows differences across the fifteen groups, the overall impression is one of surprising similarity. We calculated the correlations between all 105 pairs of cultures; the average is r = .935 with a range of .798 to .987, all p < 0.001. For all the groups except the Iranians, more than half of the statements made by negotiators were coded as information exchange—questions and answers (self‐disclosures). Also, we note that the two groups with the lowest percentages of questions, the Germans and Iranians, achieved the lowest profit levels.

Comparing the Francophone Canadians and the French shows one of the most striking similarities in verbal behavior patterns. The two rows representing observational measures of these groups’ verbal behaviors are almost identical, r = .982, p < 0.01. In contrast, both the Mexico/Spain and Anglophone Canadian/UK correlations are below the average across all groups. Perhaps these differences in “colonial/linguistic” effect are due to the greater size and diversity of the populations in Anglophone Canada and Mexico, than in Quebec.

When we scan Table Four we see impressive differences between the groups. For example, the Mexican and Spanish groups again are quite different. The Spaniards seem much more aggressive with respect to “no,” “you,” and conversational overlaps. The Brazilians are on the far end of the scale for the use of both “no” and “you.” The patterns for close East Asian neighbors Japan and South Korea are very different as well. The groups with the biggest difference in conversational overlaps are the Iranians (with the most) and the Americans (with the least).

People are generally unconscious of the differences set forth in Table Four, making them especially problematic in international negotiations. Human beings give out and take in a great deal of information nonverbally, particularly the affective sort. Differences and difficulties at this level of interaction tend to go unnoticed and unrepaired; intercultural negotiators are very often unable to articulate the reasons for their reported discomfort (cf. Lee, Yang, and Graham 2006).

Here we briefly explore associations between the observational measures (shown in Tables Three and Four) and the questionnaire data (shown in Tables One and Two). The correlations, shown in Table Five, suggest possible relationships in the data. Correlations are not causation but they do indicate relationships among variables. This quick look across methods provides a hint of systemic validity, as well as suggesting paths for future research.

Our approach in this section is a two‐stage analysis. First, individual responses are pooled by cultural group. Second, relationships between observational variables and questionnaire measures are compared. In the interests of brevity (1) we report only the statistically significant correlation coefficients, and (2) we select only a few of the more salient observational measures, including two summary measures of instrumental behaviors we call INST3neg (threats + warnings + punishments) and INST3pos (promises + recommendations + rewards). The units of analysis used here are the 15 cultural groups for which we have observational data.

It is an axiom of the literature on negotiation (cf. Lewicki, Barry, and Saunders 2016) that questions are important, and this is strongly supported by our analyses. As may be seen in Table Five below, cultural groups that used greater percentages of questions reported greater satisfaction (SATp), more of an emphasis on problem‐solving behaviors (PSAn), and greater levels of interpersonal attraction (ATTn). Groups that used higher percentages of negative instrumental behaviors (INST3neg) tended to report lower levels of satisfaction (SATp). Alternatively, groups that used more positive instrumental behaviors (INST3pos) reported lower usage of problem‐solving behaviors (PSA) and lower levels of interpersonal attraction (ATTn).

Groups that used the word “you” more frequently reported lower usage of problem‐solving behaviors (PSAn) and weaker relationships between negotiators’ PSA and profits.

(Path 3), and between negotiator attractiveness (ATTn) and partner satisfaction (SATp, Path 5).

Cultural groups that interrupted more often also tended to achieve lower profits ($n), while experiencing stronger, positive relationships between negotiator problem‐solving behaviors and partner satisfaction (Path 2). Overlaps also led to a weaker relationship between role and profits (Path 6).

In summary, the key findings are the strong associations between questions and conversational style—that is, use of the second person (“you”) and conversational overlaps—and negotiation outcomes and processes. Such insights are consistent with Lewicki, Barry, and Saunders (2016) and others who have recommended encouraging information flows in negotiations.

Now we turn to perhaps the most important question in our study. Do proven indices of cultural difference predict the variation we see in Tables One–Four above? If so, then researchers and practitioners might use scores from those well‐established indices to make predictions about the speed bumps and pitfalls in a much larger number of countries and cultures. Indeed, consider for a moment that Hofstede, Hofstede, and Minkov (2011) provided such scores for 78 countries, regions, and subcultural groups.

As in the previous section, here we employ a two‐stage analysis with individual responses pooled by country juxtaposed with a series of cultural indices. Our database includes only 20 cultural groups along 27 endogenous measures of negotiation behaviors, processes, and outcomes. For this section of the study the unit of analysis is the cultural group, so n = 20. The unprecedented scope of the data allows for comprehensive analyses of the impact of culture on marketing negotiation processes. Additionally, new measures of cultural differences are now available to support this effort.

Graham, Mintu‐Wimsat, and Rodgers (1994) compared four of Hofstede’s (2003 and with Hofstede and Bond 1988) dimensions of cultural values—power distance, individualism, masculinity, and long‐term orientation—and Hall’s (1976) low/high context concept to the same ten variables and path coefficients listed in Tables One and Two. In our current search for cultural patterns in our larger data set we initially considered a much broader array of cultural indices including Hofstede’s five dimensions of cultural values, Hall’s scheme, Schwartz’s seven dimensions (2012), the several dimensions of House et al. (2004), the importance of time (Levine 1997), linguistic distance (West and Graham 2004), and measures of trust and cultural tightness–looseness (Uz, 2014; Brett, Gunia, and Teucher 2017).

We found six indices to be most useful as indicants of patterns in our data. They are individualism/collectivism and power distance (Hofstede, Hofstede, and Minkov 2011, number of countries and regions covered, n = 78), long‐term orientation (Hofstede and Bond 1988, n = 40), the importance of time (Levine 1997, n = 32), high‐ versus low‐context communications (Hall 1976 and Meyer 2014, n = 26), and linguistic distance from English (West & Graham, 2004, n ≈ thousands!).

While there has been much criticism of Hofstede’s work over the years (cf. Kirkman, Lowe, and Gibson 2006), we continue to find that his cultural values framework demonstrates superior systemic validity with respect to several aspects of international business behavior, including linguistic distance (West and Graham 2004) and bribery (Jing and Graham 2008). Although Hofstede’s data were originally collected from 1967 to 1973, the cultural differences that he measured appear quite consistent over time.

Several researchers have pointed out that the six dimensions mentioned above are highly intercorrelated (e.g., Hofstede 2003). Cateora and Graham (2005) suggested a synthesis of these (and other) cultural indices (see Table Six below) forming a “latent construct,” one that underlies and/or subsumes the six indices. The average correlations among these six indices across our 20 cultural groups is r = .594 (p < 0.05). If we include all the scores for each variable reported in the literature and listed just above, the average correlation is r = .472 (p < 0.05).

This construct—relationship orientation—has been mentioned in other management contexts, but often with different inverses, among them, task‐, information‐ (as in Cateora and Graham 2005), arelationally‐, and rationally oriented. The concept is consistent with others’ comments about observed differences in the importance of contracts versus interpersonal relationships (Hall 1960; Graham and Herberger 1983; Salacuse 2003).

With increasing frequency, researchers in negotiation are turning their attention to relational constructs vis‐à‐vis the traditional cognitive (rational or informational) and economic aspects (Graham, Lawrence, and Requejo 2014; Salacuse 2019). Gelfand et al. (2006) theorize about the trade‐offs to be made between “economic and relational capital in negotiation.” Curhan et al. (2008) report that in two social circumstances—egalitarian contexts and female‐female dyads—negotiators tended to accommodate social relationship maintenance demands by giving up economic capital. Most recently, Unusier (2019) described the fundamental cultural difference in intercultural negotiation style to be “deal‐making” versus “relationship building.”

An important breakthrough in marketing theory during the past few decades has been the elaboration of “relationship marketing” (Gronroos 1994). Its emphasis on long‐term relationships versus transactions represents a dramatic shift in thinking about the determinants of profitability. The fundamental axiom in the area is that long‐term (i.e., loyal) customers tend to yield higher profits for firms. Meanwhile, traditional negotiation theory misconstrues interpersonal relationships as antecedents to negotiation processes and agreements (e.g., Graham, Mintu‐Wimsat, and Rodgers 1994; Lewicki, Barry, and Saunders 2016), rather than as a paramount negotiation outcome of long‐term and profitable commercial relationships. Studies of creative bargaining often focus on “creative agreements” of the narrow economic sort, while ignoring the importance of relationship maintenance (cf. Wilson and Thompson 2014; Gelfand et al. 2015).

The research designs and methods of social psychology are much to blame for this theoretical shortcoming. That is, most experimental designs ignore long‐term outcomes because they are hard to measure in the laboratory. Even when trust is considered, it is often conceptualized as an independent variable, and not as a sought‐after outcome. This problem is best exemplified by a meta‐analysis of 38 studies of trust as an antecedent (cf. Kong, Dirks, and Ferrin 2014). Our own work gives only short shrift to long‐term, relational concepts such as satisfaction and interpersonal attraction. Finally, in another field we do find empirical support for the view that interpersonal relationships must be considered along with the dominant economic constructs. In a study of pricing practices of veterinarians, Cron et al. (2009) report: “While lower prices have a negative effect on owner income, relationship orientation is found to have a positive direct effect on income … due possibly to the resulting customer loyalty.”

We expect in our 20‐country study that negotiators from more relationship‐oriented (RO) cultures, as opposed to transaction‐oriented (TO) cultures, will tend to favor more accommodating behaviors and better relational outcomes. Indeed, we see some evidence of this in the literature, as discussed above in the section containing background information. Thus our a priori expectations include the following:

1. Negotiators’ profits will be lower among the more relationship‐oriented groups.

2. RO groups will report more cooperative behavior (PSAn), greater interpersonal attraction (ATTn), and greater satisfaction (SATp).

3. Negotiator roles (Path 6, buyer or seller) will have stronger influences on profits in RO cultures.

4. Path 5, the influence of ATT on SAT, will be stronger in RO cultures.

5. Negotiators from RO cultures will use lower percentages of INST3neg behaviors, fewer interruptions, fewer “you”s, and higher percentages of questions. The last is more representative of a PSA approach.

The best approach for determining the predictive power of culture on negotiation behaviors, processes, and outcomes is to compare the latent construct of ROvTO (integrating all six cultural concepts above) to each of the endogenous variables (Tables One–Four). Because of the high intercorrelations among the six (Cronbach’s α = .80), we have modeled ROvTO as reflective indicator using a PLS (cf. Fornell and Bookstein 1982).

In Table Seven we report the results in seven columns. The first six columns include the correlation coefficients between each of the six separate cultural index elements and the endogenous variables. The last column represents the PLS path coefficients for the relationships between ROvTO and the endogenous variables.

Our findings are mostly consistent with our a priori expectations, which were based on our review of the current literature and our original 1994 article (Graham, Mintu‐Wimsat, and Rodgers 1994). That is, with better measures of cultural differences, including ROvTO, and more countries, we mostly confirm culture’s direct and moderating impacts on negotiation processes and outcomes.

Our first expectation, based primarily on the association reported in Graham, Mintu‐Wimsat, and Rodgers (1994) between negotiators’ profits ($n) and Hofstede’s individualism/collectivism index (r = .673, p < 0.05), was not confirmed using the larger sample of cultural groups. Please note that we reversed Hofstede’s IND Index for the current study, and the correlation between the same two variables reported in Table Seven was r = −.177 (p = n.s.). Also, recall that Lituchy (1997) and Arunachalam, Wall, and Chan (1998) reported that negotiators from more collectivistic cultures achieved higher profits; this is the opposite of our 1994 findings. We conclude that the best interpretation across the studies is that individualism/collectivism and negotiators’ profits are unrelated across cultural groups. Thus, we see the value of both the larger sample size and better measures in the current study. Indeed, all of the other five correlation coefficients reported in Table Seven were negative, but not statistically significant. The PLS parameter estimate for ROvTO → $n is larger and negative but also not statistically significant. Furthermore, focusing on the new data added in this study, we see that the negotiators from Japan, Brazil, Spain, and Hong Kong, all added to the data set since 1994, achieved individual profit scores among the top five of the 20 groups. Meanwhile all four scored relatively low on Hofstede’s individualism/collectivism index, thus attenuating the expected relationship between ROvTO and negotiators’ profits.

All six correlation coefficients and the path coefficient for ROvTO → PSAn (pc = .780, p < 0.01) provide strong evidence for the predictive power of culture on problem‐solving. Relationship‐oriented groups tended to use more PSAn behaviors. This result is consistent with Adler, Brahm, and Graham (1992), Graham, Mintu‐Wimsat, and Rodgers (1994), and Adair, Okumura, and Brett (2001), but contrary to Liu and Wilson (2011) and Alsani et al. (2016). Both our findings regarding the comparisons of the narrow U.S./China groups and the broader 20 cultural groups suggest skepticism about the latter studies. Also, as will be seen just below, our analyses of observational measures provide across‐methods validation of the prediction that negotiators from relationship‐oriented cultures (such as China) will tend to behave more cooperatively.

Our expectations about culture moderating the relationship between negotiators’ roles (buyer or seller) and their profits (Path 6) were unfulfilled, despite our earlier findings (Graham, Mintu‐Wimsat, and Rodgers 1994). Seven of the nine groups added to the data set since 1994 showed no moderation effects, thus attenuating the relationship reported in the 1994 study.

The moderation effect of culture on Path 5 reported in the 1994 study seems correct. The negotiators in RO cultural groups manifested more positive relationships between interpersonal attraction and satisfaction than their counterparts in TO cultural groups. All six correlation coefficients were large and positive and four were statistically significant, as was the ROvTO → $n PLS path coefficient at .513 (p < 0.05). This strong result implies that in RO cultures, the interpersonal attractiveness of negotiators will be a more important team selection criterion.

Not on our list of expectations was the discovered moderator effect of culture on the positive relationship between negotiators’ problem‐solving approach and negotiators’ profits (Path 3), long‐term orientation r = .475 (p < 0.05). That is, when negotiators used more cooperative (PSA) approaches, they were rewarded with higher profits in groups that were more long‐term oriented (Hofsted and Bond 1988).

Culture appears to affect strongly the observed negotiation behaviors in some of the ways expected. That is, negotiators from RO cultures tended to ask more questions and use fewer positive instrumental behaviors (INST3pos = promises, recommendations, and rewards) than negotiators in TO cultures. Respectively, the ROvTO path coefficients were .503 and −.463 (both p < 0.05). No relationships were found between ROvTO and the use of negative instrumental behaviors (INST3neg), the use of the word “you,” and the occurrence of conversational overlaps. Finally, Table Seven shows the overall explanatory power of two of the cultural indices, power distance (Hofstede, Hofstede, and Minkov 2011) and long‐term orientations (Hofstede and Bond 1988), as well as the usefulness of the relationship orientation latent construct.

The limitations of our work are many. A sample size of 20 tends to inflate artificially the correlation coefficients. The holes in our data exacerbate this weakness. Of course, 30 or more cultural groups with complete data would give one more confidence in our findings. It is also clear that more cultures and countries need to be studied systematically. Prominent gaps in coverage are the Netherlands (a very inventive country) and cultures in India and Africa.

It is also worth repeating that the transactional/task‐related/informational biases in the extant literature ignore the creativity of long‐term relationships as the most important outcome of negotiations. The good news is that negotiation scholars are now beginning to respond to this crucial challenge (e.g., Crotty and Brett 2012; Gelfand et al. 2015; Graham 2018).

Potential confounding factors exist in our analytical methods. For example, one reviewer rightly noted our use of individuals in a dyad as the unit of analysis. This weakness is somewhat mitigated because the members of our dyads are distinguished as buyers and sellers and we specifically model this distinction (see Kenny, Kashy, and Cook 2006). Generally our methods have been selected so as to enable us to leverage the depth and breadth of the data collected and explore for patterns in them. Our PLS approach is best suited for the task and tends to ameliorate some of the problems we have noted. Our fundamental goals are (1) to stimulate new ideas for theory building and (2) to provide systematically gathered data to better inform practitioners about the importance of cultural differences.

At the behest of a reviewer we considered two control variables. First, we found no relationship between country GDP/capita and any of our 27 endogenous variables. The data suggested that experienced businesspeople from high‐income countries negotiate no differently than those in low‐income countries.

Second, earlier we noted that we see little evidence of history effects in our data despite the 36‐year span of collection. When we looked at the correlation coefficients between year‐of‐data‐collection and the 27 endogenous variables, we found two to be statistically significant, allowing for the argument in favor of history effects: year/profits ($n), r = −.471 (p < 0.05) and year/conversational overlaps, r = .787 (p < 0.05). The −.471 suggests that over time negotiators were producing lower profits. Such an explanation defies the profusion of integrative bargaining courses and books over the 36‐year period. More likely, this effect is a consequence of the order in which we chose countries for replication, starting with major industrialized trading partners such as Japan and northern Europe.

We also compared profits ($n) over the seven waves of data collection of the 160 American participants. For the 1979 group (n = 38) the average profits were 44.6 and for the 1987–1991 groups (total n = 122) the average was 45.0. The difference is neither practically nor statistically significant over the decade of data collection. Indeed, the means for the seven waves were, in order, 44.6, 43.3, 44.9, 44.9, 43.7, 46.6, 44.0, representing remarkable consistency.

We attribute the difference in conversational overlaps to the order of countries chosen—the American and Japanese data collected in 1979 had the fewest number and the Iranian data collected in 2015 had the highest number.

This brief discussion about history effects, and the lack thereof, carries with it an important implication. Negotiation style appears very sticky (as in the economic concept of “sticky” prices, i.e., durable) over time. The continuing value of Hofstede’s work—his indices are based on data collected in 1967–1973—supports the idea that culture can be very durable indeed. A very interesting future research opportunity would be a replication of the study in Russia. Has the dramatic change in politics there affected negotiation behaviors, processes, and outcomes? Even more interesting will be replications here in the United States with a newer generation of businesspeople. Turkle (2016), among many others, claims that conversational style in the United States is undergoing fundamental change because of the proliferation of smart phones. Most recently, we might anticipate a COVID‐19 effect on local and global interaction as well.

We are confident about the validity of our findings regarding the influence of cultural values on negotiation behaviors. However, we are just scratching the surface of the rich and important phenomena of emotions and nonverbal behaviors in negotiation. Research has long shown facial expression to be more important than any other channel of interpersonal communication (Mehabrian 1972; Ekman 1973). We now have machine‐based means to record and code facial expression data (Tiexeira, Picard, and el Kaliouby 2014). Therein lies the greatest opportunity for new research in this area.

This report is data rich. Indeed, we hope that other researchers will develop other approaches for analyzing these data (and associated data that we will provide upon request). Also, we encourage other researchers to replicate our studies in other regions of the world and share their findings.

The primary strength of this research is the breadth and depth of the data collection. There are no other studies in international negotiation that involve experienced businesspeople from such a large array of cultures and countries. Consistent with Graham, Mintu‐Wimsat, and Rodgers (1994), the participants’ cultural diversity yields a wonderful complexity in approaches to a fundamental human behavior, that is, negotiation.

The rich variation in negotiation rituals across the 20 cultural groups included in this study is evident. Even so, the fundamental causal paths (Figure One) underpinning these analyses seems correct: culture affects negotiation behaviors, processes, and outcomes and moderates relationships among them. Our findings are consistent with previous pathbreaking studies finding that moderation effects are conceptually and empirically important (cf. Liu et al. 2012; Gelfand et al. 2015).

Our work makes several contributions to negotiation studies. First, we provided new data on the negotiation behaviors, processes, and outcomes of five countries and regions: southern China, Hong Kong, the Czech Republic, the Philippines, and Iran.. We combined this data with data from previous studies, bringing to 20 the number of cultural groups analyzed. This allows a stronger platform for statistical tests of relationships. We are able to apply better measures (i.e., having more reliability and greater validity) of cultural differences. These data and measures are useful in our analyses. Moreover, the data themselves can help negotiators to anticipate the degree and kinds of cultural differences that might be expected across 190 types of cross‐cultural dyads.

A second contribution is our comparison of questionnaire measures and observational measures of negotiation behaviors and outcomes. See Table Five. These comparisons both broaden our appreciation of the value of observational methods and add to the systemic validity of some of our measures, analyses, and theories. For example, cultural groups that asked higher percentages of questions achieved higher levels of satisfaction and interpersonal attraction. And, as we would expect, questions also led to higher scores on our PSA scale. Alternatively, higher percentages of instrumental behaviors (e.g., promises and threats) reduced both satisfaction and attraction. Conversational overlaps appear to have directly reduced negotiators’ profits. A key finding with respect to the content analyses is the greater differences across the nonverbal and linguistic style variables than the verbal behaviors.

Third, we have derived a new latent construct of cultural difference, something we call relationship‐oriented versus transaction‐oriented (ROvTO) cultural groups. This construct, borrowed from relationship marketing theory, was useful in summarizing cultural variation, while focusing on those behaviors most clearly related to negotiation phenomena.

But perhaps the greatest value of this new construct is that it starkly reveals a general criticism of the literature—the focus on agreements, rational behaviors, and short‐term outcomes has greatly limited our understanding of how humans interact in international commerce. Our own work suffers from this limitation. This “bounded rationality” is a consequence of both an American scientific ethnocentricity and an associated premature experimental‐design bias in the literature (cf. Graham and Gronhaug 1989). Simply put, negotiation is not a competitive game.

We note the centrality of the concept of long‐term orientation (Hofstede and Bond 1988) and its direct and moderating effects on negotiation behaviors, processes, and outcomes (see Table Seven). This finding argues for longitudinal research designs such as that of Watson, Kumar, and Michaelsen (1993).

We also note the potential value of linguistic distance from English (West and Graham 2004) as a potential proxy for relationship orientation. The measure is strongly intercorrelated with the other five cultural indices listed in Table Six and can be applied to some 7,000 languages. Its best attribute is its ease of measurement for groups and/or individuals; one need ask the simple question, “What is your native tongue?” Of course, there are issues related to bi‐ and multilingual research participants, but those can be sorted out in future studies.

We can discern patterns in the data that recommend both future confirmation and new directions in research methods and theories. Cultural groups that place less value on time achieve higher levels of both interpersonal attraction and satisfaction. Cultural groups that are more relationship‐oriented ask more questions and use more problem‐solving behaviors. We learn that in more relationship‐oriented cultural groups, negotiators that take a problem‐solving approach achieve higher economic profits. Also, in more relationship‐oriented groups there is evidence that interpersonal attraction has a stronger impact on satisfaction.

We began this article with a brief comment on differences in buyer–seller interactions in the United States and Japan. Invention best happens in the context of long‐term, cooperative relationships. Although the Japanese buyer–seller relationship is clearly hierarchical and hierarchy generally kills creativity, the Japanese negotiators produced the highest joint profits among the 20 cultural groups in Kelley’s (1966) mixed‐motive negotiation simulation (see Tables One and Two). The assumed context in Japan is long‐term commercial and inventive relationships that promote creativity. While the American commercial approach emphasizes efficiency through competition, the Japanese approach accomplishes similar goals through reducing transaction costs (Hodgson, Sano, and Graham 2018).

This is also the lesson of the literature on relationship marketing. Such reasoning views interpersonal attractiveness as an outcome of negotiations rather than an antecedent. This leads to the question of how we can develop and include measures of creativity and long‐term relationships in negotiation research. There are useful hints in the work of Watson, Kumar, and Michaelsen (1993), which reports that diverse working groups initially produced fewer new ideas than comparable homogeneous groups, but in the long run, the diverse groups were more productive.

The metaphors or frameworks of competitive games or even integrative problem solving, which we have used here in developing our database, do not well reflect the reality of international commerce, in which long‐term relationships, invention, and the exploitation of mutual opportunities predominate—as long as businesspeople can see and avoid the speed bumps and pitfalls in their international negotiations. We provide help in this last regard in our article on the practice of international relations, forthcoming in Negotiation Journal 36(3) (fall 2020).