INTRODUCTION
Cross-cultural studies are a niche area in experimental economics that has made considerable advances in the past two decades in terms of better understanding the influence of culture on economic decision-making (van Witteloostuijn, Reference Van Witteloostuijn2015). The widely believed advantage of experimentation is that it provides empirical results based on decisions that are directly affected by monetary rewards (incentives). Hence, incentivized experiments are believed to trigger intrinsically motivated behavior to a much higher extent than, for instance, questionnaire-based surveys would do (Roth, Reference Roth, Kagel and Roth1995; Smith & Walker, Reference Smith and Walker1993). A further advantage is that factors that influence decision-making behavior can be controlled and the data gathering itself can be replicated at the same or in different places (labs) or by using subject pools from different countries (Binmore & Shaked, Reference Binmore and Shaked2010; Gächter, Reference Gächter2009; Ottone, Ortona, Ponzano, & Scacciati, Reference Ottone, Ortona, Ponzano and Scacciati2010; Weber & Camerer, Reference Weber and Camerer2006). By observing the decision-making behavior of test persons from different countries who are tested under the same conditions, observed behavioral differences are often attributed to the influence of cultural differences. However, although the field has progressed, as we will see in the course of this paper, nevertheless cross-cultural experimental research in economics still faces limits of explanatory depth to explain why culturally-induced behavioral differences occur. Conventionally, behavioral differences are often interpreted en bloc as cultural differences. As a result, they explain little about the distinctive characteristics that lead to the observed differences. In recent cross-cultural experimental economics (CCEE), a call for a deeper understanding of the factors that has led to those differences has arisen (Chuah, Hoffmann, Jones, & Williams, Reference Chuah, Hoffmann, Jones and Williams2007; Oosterbeek, Sloof, & Van De Kuilen, Reference Oosterbeek, Sloof and Van De Kuilen2004).
Whereas van Witteloostuijn (Reference Van Witteloostuijn2015) provides convincing reasons for the application of experiments in the field of International Business (IB), this study focuses more on the micro-level of human behavior relevant for cross-cultural research in the field of international management, particularly indigenous management research. The purpose of this study is first and foremost to introduce indigenous management scholars to studies conducted in the field of CCEE. I do not aim to produce a complete literature review of the study of culture in the economic discipline. Rather I intend to present experimental applications and insight and outline methodological advantages that I propose to merge with traditional approaches to indigenous management research in order to propose advancement from a methodological point of view. By doing so, I first aim to generate an understanding about the state of research in the field of CCEE by focusing on the question of what CCEE has contributed to the understanding of culture. Further, based on the latter, I debate how Indigenous Management Research (IMR) and CCEE can mutually benefit by combining methodological strengths for the study of culture. I approach this question by systematically reviewing studies published in this field. Further, the results are summarized and critically discussed, and paths are proposed for advancement.
LITERATURE SELECTION AND CATEGORIZATION
As this paper intends to review, categorize, and integrate what has been achieved in the niche of CCEE so far, articles qualify if they fulfill the following criteria. First, they must be relevant to the field of experimental economics (i.e., by applying research methods common in this discipline). Second, they must be relevant to cross-cultural economic research, namely only those articles that explicitly aim to examine culture were considered. Third, the research should have been published in an internationally recognized peer-reviewed journal.
A keyword search was undertaken in order to identify relevant articles in the first round of filtering. The following keywords were used: experiment, culture, cross-culture, inter-culture, values, norms, behavior, and differences. The sources used for the keyword search were Google Scholar, EconLit, Web of Knowledge, and Ideas. This search resulted in numerous relevant studies that partly included non-experimental studies or studies that did not show a direct link with the examination of cultural phenomena. In the second round of filtering, only those studies that apply an experimental methodology for the examination of culture were selected. As a result, 23 studies were selected that were published in journals including the American Economic Review, Journal of Economic Behavior & Organization, Experimental Economics, Journal of Socio-Economics, or Journal of Public Economics and Science (see Table 1).
Rank (#) according to citations of selected studies
Source: Google Scholar (3 May 2013) and Publish or Perish (3 May 2013; Harzing, Reference Harzing2007)
Following the selection criteria described above, relevant studies were analyzed in two steps. In the first step, articles were clustered according to research design (i.e., by game application). In the second step, within each cluster, studies were sorted according to the subject of analysis, enabling the identification of the broad research streams within each cluster.
However, while operating the structured approach described above, I faced the limitations that it caused. For instance, as is conventionally done in management studies, I initially followed the popular and common ‘Hofstedian’ interpretation of culture, which uses country culture foremost as a proxy for culture. While this is common, it is blind on newer developments that pursue a more realistic view on culture and includes – as advanced countries become more multicultural – race, ethnicity, or religion as anchors of social identity. Thus, I complemented my approach by presenting studies that reflect newer trends and that have meaning for cross-cultural research but that were not adequately captured by our criteria. Additionally, I report about the few experimental studies conducted in management studies (Leung, Bhagat, Buchan, Erez, & Gibson, Reference Leung, Bhagat, Buchan, Erez and Gibson2005; Leung & Su, Reference Leung, Su, Punnett and Shenkar2004).
Further, I have added to the table information on how far the respective studies discuss questions of size effects, and where applicable null and negative findings, since this is today a common standard contrary to the past where the unexplained variance in the data was overwhelming at times and left doubts in terms of the explanatory power of this research (Lewin et al., Reference Lewin, Chiu, Fey, Levine, McDermott, Murmann and Tsang2016). Benchmark studies that explicitly and comprehensively discuss size effects are the studies by, e.g., Chua and Ng (Reference Chua and Ng2017) and in particular Gerhart (Reference Gerhart2009) on culture-related research. In order to bolster the explanatory power of the empirical findings, studies should inform how high or low the variance is and how this can be explained; discuss null and negative finding as well as to highlight the practical importance of the results. However, since especially in the field of CCEE several studies are rather explorative in nature, i.e., not guided by a set of hypotheses (e.g., Burns, Reference Burns2016; Chuah et al., Reference Chuah, Hoffmann, Jones and Williams2007; Chuah, Hoffmann, Ramasamy, & Tan, Reference Chuah, Hoffmann, Ramasamy and Tan2014; Henrich et al., Reference Henrich, Boyd, Bowles, Camerer, Fehr, Gintis and McElreath2001, Reference Henrich, McElreath, Barr, Ensminger, Barrett, Bolyanatz and Ziker2006, Reference Henrich, Ensminger, McElreath, Barr, Barrett, Bolyanatz, Cardenas, Gurven, Gwako, Henrich, Lesorogol, Marlowe, Tracer and Ziker2010), in many cases they are not amenable to discuss null and negative findings.
In the following subsection, by starting with the CCEE literature, the result of each study in the respective category is presented in the form of an overview. Since the studies that I present can be regarded as the ‘first major wave’ of CCEE studies, they can be allocated to the decade of the early 1990s and beginning of 2000s. In that era, strategic games were most common for studying cross-cultural phenomena. I note that I provide further explanations on the most common game designs and rules in Appendices I–IV, as well as other game designs I came across in Appendix V. Moreover, I point out other applied designs in the text. The following section is succeeded by a review of newer trends in the CCEE field.
Fairness and Context
Fundamental research about fairness motives in different contexts (e.g., market versus dyadic) has been conducted, among others, by Kahneman, Knetsch, and Thaler (Reference Kahneman, Knetsch and Thaler1986), Rabin (Reference Rabin1993), or Fehr and Schmidt (Reference Fehr, Kirchsteiger and Riedl1993). The main motivation of these early studies was first and foremost a discomfort with the very central assumption of economic behavior that proposes that ‘almost all economic models assume that all people are exclusively pursuing their material self-interest and do not care about “social” goals per se’ (Fehr & Schmidt, Reference Fehr and Schmidt1999: 817). Experimental studies showed that in fact social goals do play a role, which goes against the basic assumptions manifested in the model of the homo oeconomicus. In a departure from there, scholars developed models of social preferences explicitly assuming both self-interest as well as a concern for payoffs of others (Charness & Rabin, Reference Charness and Rabin2002; Fehr & Gaechter, Reference Fehr and Gächter2000). As a result of this work, theories have been developed further that make a case for human reciprocity behavior resulting in the development of a new species, the homo reciprocans (Bowles & Gintis, Reference Bowles and Gintis2002; Fehr & Gaechter, Reference Fehr and Gächter1998). In parallel to these developments, scholars became interested in the question of whether cultural factors may play a role, too. Culture-exploring studies based on the popular ultimatum game (UG, for a detailed description, see Appendix I) have been conducted by Buchan and Croson (Reference Buchan and Croson2004), Chen and Tang (Reference Chen and Tang2009), Chuah et al. (Reference Chuah, Hoffmann, Jones and Williams2007, Reference Chuah, Hoffmann, Jones and Williams2009), Hennig-Schmidt, Li, and Yang (Reference Hennig-Schmidt, Li and Yang2008), Henrich et al. (Reference Henrich, Boyd, Bowles, Camerer, Fehr, Gintis and McElreath2001), Roth, Prasnikar, Okuno-Fujiwara, and Zamir (Reference Roth, Prasnikar, Okuno-Fujiwara and Zamir1991), and Valenzuela, Srivastava, and Lee (Reference Valenzuela, Srivastava and Lee2005). These studies can be categorized according to three groups: (1) those that detect cultural differences with regard to norms of fairness, (2) those that analyze differences in fair beliefs, and (3) those that include further contextual factors that influence the decision-making behavior of individuals and groups in the inter- and cross-cultural contexts (Table 2).
Cross-cultural experimental studies based on Ultimatum Game (UG) applications (for a detailed description incl. game tree, see Appendix I)
The seminal work conducted by Roth et al. (Reference Roth, Prasnikar, Okuno-Fujiwara and Zamir1991) and Henrich et al. (Reference Henrich, Boyd, Bowles, Camerer, Fehr, Gintis and McElreath2001) tests the rational-behavior assumption of homo oeconomicus and examines economic behavior in 15 (mostly tribal) societies (Henrich et al., Reference Henrich, Boyd, Bowles, Camerer, Fehr, Gintis and McElreath2001). The standard behavioral assumption of the homo oeconomicus could not be confirmed in any of the societies investigated as considerable differences in behavior between the respective cultures were observed. The participants of some tribes offered 25% or more of the stakes at play, while others offered even more than 50% of the stake. Roth et al. (Reference Roth, Prasnikar, Okuno-Fujiwara and Zamir1991) use a subject pool from Israel, Japan, the US, and the former Yugoslavia. Their results show that the observed differences, such as high offers made in the US and former Yugoslavia compared with the low offers made in Israel, cannot be explained by language, currency, or experimental design effects, which are considered to distort behavior (Roth, Reference Roth, Kagel and Roth1995), but rather by cultural differences. They conclude that behavioral differences seem to be the result of varying fairness norms.
Buchan and Croson (Reference Buchan and Croson2004) focused on the influence of fairness concepts on negotiation behavior. In their study, the repeated UG was played in Japan and the US with fairness norms differing significantly in the countries analyzed. Americans believe that the actor with greater negotiation power should be allocated the highest payment, while the Japanese expect the actor with greater negotiation power to share the payment with the weaker actor. Further, a religious mindset may influence fairness concepts and thus differences in decision-making behavior. Chen and Tang (Reference Chen and Tang2009) apply the UG in order to analyze the effect of cultural factors on behavior, with particular regard to beliefs. Tibetans in Lhasa and Han Chinese in Xiamen as well as in Singapore were recruited from groups of pupils, students, and professors. While the behavior of the Han Chinese was internally homogeneous, Tibetans demonstrated a higher probability of accepting offers per se, which remained largely unaffected by the amount offered. The authors ascribe this result to the cultural differences explained by Tibetans’ religiousness.
In addition to differences in fairness concepts, the group context and location in which decisions are made influences the outcome and reflects the different levels of social orientation across cultures. For instance, Valenzuela et al. (Reference Valenzuela, Srivastava and Lee2005) examine the effect of different cultural orientations by means of an UG with an incomplete distribution of information in the US and South Korea. In comparison to Americans, South Korean test persons show a distinct sensitivity to the influences of the group as well as to the situational context as soon as information about them was available. The authors assume that Korean participants are more likely to accept a relatively low offer when decisions are made that include the group context. In the US, however, no such effect is observed. Chuah et al. (Reference Chuah, Hoffmann, Jones and Williams2007, Reference Chuah, Hoffmann, Jones and Williams2009) compare the decision-making behavior of participants from Malaysia and Great Britain by taking the decision-making location into account. Malay proposers and responders in Malaysia and British proposers and responders in Great Britain show significant differences in proposer behavior. The amount offered is generally higher in the Malay treatment group. Equally interesting is the result of the inter-cultural scenario. In this case, Malays generally offer lower amounts to the British but not to other Malays. The British, however, do not change their behavior. The results significantly testify that group homogeneity and location (country) affect bargaining behavior, i.e., when Malays played in Britain, their offers were higher.
Cooperation and Sanctioning
Representative studies that examine cultural factors in public good (PG) games (for a detailed description, see Appendix II) have been conducted by Weimann (Reference Weimann1994), Burlando and Hey (Reference Burlando and Hey1997), Cason, Saijo, and Yamato (Reference Cason, Saijo and Yamato2002), Castro (Reference Castro2008), Kocher, Cherry, Kroll, Netzer, and Sutter (Reference Kocher, Cherry, Kroll, Netzer and Sutter2008), Herrmann, Thöni, and Gächter (Reference Herrmann, Thöni and Gächter2008), and Gächter and Herrmann (Reference Gächter and Herrmann2009). Table 3 provides an overview of the studies analyzed.
Cross-cultural experimental studies based on Public Good (PG) game applications (for a detailed description incl. game tree, see Appendix II)
Notes: * Burlando/ Hey (Reference Burlando and Hey1997) and Weimann (Reference Weimann1994) refer in their investigations to the results of the Andreoni experiments in which the behavior of American students was examined based on a PG game.
Central to the public good game is the conventional rationality assumption of economic behavior that predicts that the contribution of every player equals zero regardless of the number of rounds played. In experiments, however, this assumption cannot be verified. Participants instead contribute approximately 50% of their lot during the first round, which, however, decreases with the number of rounds. No contribution is an exception (Weimann, Reference Weimann1994). Applying the public good game in a cross-cultural setting sheds more light on the test person's willingness to cooperate, the structure of interaction, and social behavior. Weimann (Reference Weimann1994) explores whether the offer levels contributed during the game differ between strangers or partners. Subsequently, he refines Andreoni's (Reference Andreoni1988) study, which concludes that strangers generally contribute more, but that the contribution amounts of both strangers and partners drop to a minimum during the last round. Andreoni assumes that this may be the result of cooperative efforts to establish social norms, which, however, do not prevail. The declining amount is therefore an expression of capitulation. However, Weimann (Reference Weimann1994) could not reproduce Andreoni's results in the sense that strangers always contribute more. Andreoni conducted the experiment with US students and Weimann with German students. Just as Andreoni assumes the possibility of cultural influences, Weimann interprets his finding such that American students act more rationally than German ones. Further, Burlando and Hey (Reference Burlando and Hey1997) also repeat the Andreoni experiment using British and Italian subject pools. Their result concurs with Weimann's findings. Although the focus of their study adopts Weimann's assumption of cultural differences, Burlando and Hey (Reference Burlando and Hey1997) specifically find that Anglo-Saxons tend to free-ride compared with Italians: the tendency to free-ride shows a higher magnitude for British than for Italians.
Castro (Reference Castro2008) examines the cultural differences between Italian and British students using the PG game. Participants are divided into homogeneous and mixed groups. The result shows that British students contribute more to the public good than Italian students. From a cultural perspective, the conclusion of the study is interesting as it states that the contributions in the culturally homogeneous groups are higher than those in the mixed groups. Kocher et al. (Reference Kocher, Cherry, Kroll, Netzer and Sutter2008) explore the extent of conditional cooperation in the US, Austria, and Japan. Conditional cooperation is understood as the will to contribute to a PG under the premise that others contribute as well. The authors conclude that both the frequency of cooperation and the relation between one's own contribution and those of other participants are significantly higher among US participants compared with the behavior observed in Austria and Japan. The latter finding implies that the inclination to contribute to a public good, if others contribute as well, is fairly high among American test persons. The higher the contributions of other actors, the higher are their own contributions.
The actions taken by actors to establish social norms can be examined as an aspect of cooperation behavior. The sanctioning of players is an opportunity to effect a change in behavior. Cason et al. (Reference Cason, Saijo and Yamato2002) explore spiteful behavior and cultural differences within the framework of a two-person PG game with US and Japanese test persons. Participants have, in a departure from the conventional modeling of a PG game, the opportunity to tell the other players if they will contribute to the PG or not before they make their decisions. Therefore, players are given the possibility of taking this information into account when making their own decisions. Spitefulness is understood as a sanctioning mechanism toward other members who only contribute very little to the PG. This manifest itself as comparably low contribution amounts among Japanese test persons as soon as it is known that one player will make no contribution. This behavior leads, among Japanese test persons, to a more efficient overall result. Whereas American participants contribute higher amounts in general, cooperation is lower when a player makes no contribution. As a result, the authors conclude that Japanese test persons tend toward spiteful behavior toward free-riders expressed by low contributions. Further, Gächter and Herrmann (Reference Gächter and Herrmann2009) test participants from Switzerland and Russia. In groups of three, a one-shot PG game is played in two variants, namely one option that provides the opportunity to sanction and one that does not. No difference in sanctioning is observed within the context of the respective group. However, both groups show significant differences: the sanctioning in the Russian group turns out to be more severe than that in the Swiss group. Not only is free-riding and the contribution of small amounts punished but so is the contribution of amounts that lay within the group average. The latter is described as antisocial reciprocity (punishment) and applies to the punishment of absolutely positive contributions. The authors suggest that the cultural determinants of negative reciprocity should be investigated further in order to examine why punishment also occurs with pro-social behavior.
By using an identical research design, Herrmann et al. (Reference Herrmann, Thöni and Gächter2008) play a PG game in 16 countries with the goal of examining differences in sanctioning behavior. The results again show strongly divergent patterns of behavior between the participants from the respective countries. Whereas in some countries, very low contributions are punished, higher-than-average contributions are punished in other countries. The authors term this form of behavior antisocial punishment. Antisocial punishment, observed in all countries, can be, according to the authors, a form of revenge or an instrument to force cooperation. The respective distinctions, however, can be a sign of culturally different personality traits such as the striving for dominance of a competitive personality, or the pursuit of maximum advantage. On the other hand, those who contributed little and who were punished by those who contributed a lot may feel exposed and thus punish them in return. A further explanation would be striving for normative equality, according to which all dissenters from the norm, whether due to very little or very high contributions, experience punishment.
Corruption and Social Preferences
Cross-cultural studies applying the trust game (TG, for a detailed description, see Appendix III) include Carpenter, Daniere, and Takahashi (Reference Carpenter, Daniere and Takahashi2004), Buchan and Croson (Reference Buchan and Croson2004), Buchan, Johnson, and Croson (Reference Buchan, Johnson and Croson2006), Bohnet, Greig, Herrmann, and Zeckhauser (Reference Bohnet, Greig, Herrmann and Zeckhauser2008), Cameron, Chaudhun, Erkal, and Gangadharan (Reference Cameron, Chaudhuri, Erkal and Gangadharan2009), and Barr and Serra (Reference Barr and Serra2010). They can be further classified into studies that examine social preferences in combination with the attribution of trust and the cross-cultural analyses of corrupt behavior (Table 4).
Cross-cultural experimental studies based on Trust Game (TG) applications (for a detailed description incl. game tree, see Appendix III)
Notes: ** In order to measure fraud aversion as well as risk behavior, a TG and a dictator game are adopted. *** The investment game is a variant of the TG.
Carpenter et al. (Reference Carpenter, Daniere and Takahashi2004) examine the factors of trust and cooperation among urban slum inhabitants in Thailand and Vietnam. Their data indicate that culture plays a major role in the extent to which people trust others and cooperate. As a result, the authors assume that cooperation behavior is determined by demographic factors such as gender, education, age, and affiliation time within a culturally homogeneous group. Based on the early studies of Hall (Reference Hall1959), who concludes that in some cultures trust is more highly valued than binding contracts, Buchan and Croson (Reference Buchan and Croson2004) compare the quality of trust in the US and China. In their study, they make use of a variant of a TG, namely an investment game (Berg, Dickhaut, & McCabe, Reference Berg, Dickhaut and McCabe1995). The peculiarity of trust is, at a general level, comparatively higher within the Chinese ingroup compared with the US one. However, at a more detailed level the results are rather heterogenic, leading Buchan and Croson to argue that ‘the importance of trust (and trustworthiness) to economic relationships and growth suggests that the issue of national differences in trust and trustworthiness is one that we as economists need to better understand’ (Reference Buchan and Croson2004: 498).
Buchan et al. (Reference Buchan, Johnson and Croson2006) examine the influence of social distance and communication on trust and reciprocity using participants from China, Korea, Japan, and the US. After randomly selecting discussion groups, player pairs are composed with the goal of creating in- and outgroups. Some player pairs are composed of people who knew each other from the prior discussion (ingroup), while the other player pairs are composed of people who had no previous contact (outgroup). The authors testify that having previous personal contact leads to significantly higher trust ascriptions, even although no game strategies were agreed upon during the personal discussion. Contrary to the assumption that Chinese participants might show preferences toward their compatriots, they show more trust and reciprocity toward the outgroup, while American test persons show the contrary behavior. A survey conducted to analyze this result attests that Chinese participants have a high level of collectivistic attitude, while this does not especially differ among the other participants (those from the US, Japan, and Korea).
Bohnet et al. (Reference Bohnet, Greig, Herrmann and Zeckhauser2008) analyze the level of betrayal aversion by employing a TG, a decision game, and a dictator game with test persons from Brazil, China, Switzerland, Turkey, the United Arab Emirates (UAE), and the US. The study asks how far the level of betrayal aversion is influenced by another person (social risk) or by a higher force. The authors use a risk factor to embody the minimal probability of the acceptance of the test persons within a risk situation that can lead to the exposure of betrayal. Within the group of the Brazilian test persons, no signs of betrayal aversion could be detected, whereas this was especially distinct among the participants from the US and UAE. This finding can be attributed to the existence of specific (cultural) institutions in both countries that counteract fraudulent intentions and thus keep the material costs of betrayal rather low. In the US, this kind of betrayal-intention-minimizing institution is embodied by the legislative body, while in the UAE these institutions are mainly those influenced by Islamic culture as well as interpersonal relations.
Cameron et al. (Reference Cameron, Chaudhuri, Erkal and Gangadharan2009) analyze cultural differences to examine tolerance toward corruption in Australia, India, Indonesia, and Singapore. The authors found that Indian participants show a higher tolerance toward corruption than participants from Australia. Contrary to expectations, participants from Singapore show a higher tolerance toward corruption and participants from Indonesia a far lower one. The perception of corruption costs thus seems to be culture-specific. By using an identical research design, Barr and Serra (Reference Barr and Serra2010) included a further stage in which a player has the choice to punish another player for the acceptance of a bribe. The game consists of three players who each play a one-shot game in three rounds. The first player represents a company representative who decides to offer a sum to a public official. He or she in turn can decide to accept or refuse the sum. The third player, who embodies a citizen, has the opportunity to punish him or her if he or she accepts the sum. Participants are recruited from 40 countries located either very low or very high on the Corruption Perception Index. The results of a repeat inquiry two years later led the authors to the finding that the inclination to behave corruptly correlated negatively with the duration of stay in Great Britain: the longer the students remained in Great Britain, the less they were inclined to act corruptly. Finally, the authors suggest that individual corrupt behavior can be regarded as a cultural phenomenon that is influenced by the environment.
Norm Enforcement, Punishment, and Inequality Aversion
Research designs that include more than one game application, mixed games (MG), allow for a broader perspective of analyzing decision-making behavior (Fehr & Fischbacher, Reference Fehr and Fischbacher2004). Although cross-cultural economic experiments using an MG approach are still rare, three remarkable studies are presented in the following that use an identical game mix that consists of an ultimatum game, dictator game, and third-party punishment game (for a detailed description, see Appendix IV). The dictator game can be regarded as a variant of the ultimatum game, whereby the feedback option is not given to the responder (see also Appendix V). Only the proposer makes the decision about the share to be allocated to the receiver, which indicates the measure of fairness. Cultural differences were investigated in the framework of an MG design by Henrich et al. (Reference Henrich, McElreath, Barr, Ensminger, Barrett, Bolyanatz and Ziker2006), Barr et al. (Reference Barr, Wallace, Ensminger, Henrich, Barrett, Bolyanatz, Cardenas, Gurven, Gwako, Lesorogol, Marlowe, McElreath, Tracer and Ziker2009), and Henrich, Hein, and Norenzayan (Reference Henrich, Heine and Norenzayan2010). These studies are summarized in Table 5.
Cross-cultural experimental studies based on Mixed Game (MG) applications (for a detailed description incl. game tree, see Appendix IV)
Note: * Partly overlapping with Henrich et al. (Reference Henrich, McElreath, Barr, Ensminger, Barrett, Bolyanatz and Ziker2006).
Henrich et al. (Reference Henrich, McElreath, Barr, Ensminger, Barrett, Bolyanatz and Ziker2006) use a non-student subject pool that includes participants from 15 tribal populations from Africa, North and South America, Asia, and Oceania. Broadly formulated, the authors study cooperation behavior, specifically focusing on the (high-cost) punishment of other participants for unequal behavior. The results across the subject pools show that people are willing to exert costly punishment but that the willingness to punish differs across societies. Moreover, some communities punish very low offers as well as offers that are very high. In addition to these findings, the authors show that communities that have a high willingness to exert costly punishment also show more altruistic behavior. In a follow-up study, Henrich et al. (Reference Henrich, Heine and Norenzayan2010) investigate the evolution of fairness and punishment by using a partly overlapping subject pool from the survey in 2006. These findings can be regarded as an extension of those derived from the prior project. The authors find that the observed tribal communities, characterized by a small community size and low degree of market integration, show little motivation toward punishing unfair offers and are generally less concerned with fair behavior per se. By comparison, the larger the observed communities are, the higher is the willingness to engage in punishment. Whereas in smaller societies kinship- and reciprocity-based norms prevail, Henrich et al. (Reference Henrich, Heine and Norenzayan2010b) summarize that the evolution of social norms in complex societies is influenced by those norms that ensure fair transactions.
Barr et al. (Reference Barr, Wallace, Ensminger, Henrich, Barrett, Bolyanatz, Cardenas, Gurven, Gwako, Lesorogol, Marlowe, McElreath, Tracer and Ziker2009) inquire whether differences in inequality aversion can be explained by behavioral diversity. Participants in this study comprise students from the US and hunter-gatherer tribes from the Amazonia, the Arctic, and Africa. The authors apply an MG design in the fashion of that used by Henrich et al. (Reference Henrich, McElreath, Barr, Ensminger, Barrett, Bolyanatz and Ziker2006, Reference Henrich, Heine and Norenzayan2010b) in order to study the participants’ appreciation of equality. In order to measure inequality aversion, a U-shaped utility function is used that considers the rejection of low offers (negative reciprocity) as well as the rejection of high offers. The authors find significant differences in decision-making behavior in all three games across all communities based on valuing equality; in other words, inequality aversion differs across countries. A universal attitude toward equality does not exist.
New Trends – Moving Beyond National Culture as a Proxy Variable
By extending the definition of culture away from the sole country-culture view, the field of CCEE has recently progressed by going one step further and trying to unpack culture through focusing on (shared) values that may be responsible for behavioral differences, or by exploring the manifestation of social identity and discrimination when people from different cultures interact. In addition, several recent CCEE studies analyze ethnicity, race, and religion as proxies for cross-cultural differences.
Diverging sets of values
Since approximately the second half of the first decade of the 2000s, cross-cultural experimental economists have moved on from merely recognizing behavioral differences between country-specific subject pools to trying to explain why these differences occur. One way to ‘unpack’ culture is by approaching it with a focus on values, as done by Chuah et al. (Reference Chuah, Hoffmann, Jones and Williams2009), who used an experimental approach in connection to attitudinal surveying in order to analyze bargaining behavior of Malaysian and British test persons. Dimensions of the World Values Survey were utilized in order to explain differences in attitudes, interpreted as cultural differences that may explain differences in decision-making behavior. Applying a bargaining game, the results showed that Malaysian proposers made higher offers to compatriots, but British proposers did not change their offer behavior based on responder nationality. While both groups differ in a variety of dimensions of the World Values Survey, it is interesting to note that the Malaysian test persons held more negative views toward outgroup members and were less in favor of expressing individual freedom and women's equal rights. Based on a regression analysis of the results, it is suggested that attitudes rather than nationality are a more powerful predictor of behavioral differences.
By using a subject pool consisting of test persons from Finland, Lönnqvist, Walkowitz, Wichardt, Lindeman, and Verkasalo (Reference Lönnqvist, Walkowitz, Wichardt, Lindeman and Verkasalo2009) show that a person's conformism values are an antecedent of whether behavior is guided by an individual's values or by social norms. Three separate studies that the research includes come to the conclusion that conformism values moderate the influence of group norms on behavior. Further, in a group of people scoring low in conformism values, universalism values (Schwartz, Melech, Lehmann, Burgess, Harris, & Owen, Reference Schwartz, Melech, Lehmann, Burgess, Harris and Owens2001) strongly predicted altruistic behavior.
Today, one stream of CCEE research progressed by focusing more on studying the influence of values on behavior by treating values as a proxy instead of national culture (see also Lönnqvist , Walkowitz, Verkasalo, & Wichardt, Reference Lönnqvist, Walkowitz, Verkasalo and Wichardt2011). A further advancement can be seen in the research stream on ethnicity, race, and religion as focus of research. The more countries become multicultural, the more difficult it is to think about culture only in the conventional dimensional categories. A more recent research stream treats, for instance, ethnicity, race, and religion as proxies for culture.
Ethnicity, race, and religion
Chuah et al. (Reference Chuah, Hoffmann, Ramasamy and Tan2014) explore cross-ethnicity with test persons from Malaysia, a multicultural country composed of four main ethnic groups with differing languages and religious beliefs. The research design featured an experiment and a questionnaire to capture the subjects’ demographics, ethnicity, religiosity, and fundamentalism. The study explores how the latter affect cooperation. Overall, ethnic and religious identity increases cooperation behavior; although religions identity increases it to a greater extent than ethnicity. Moreover, the sole knowledge of ethnic and religious differences does not influence cooperation, but the awareness of it increases cooperation. Also, a person's religiosity and fundamentalism do not influence cooperation. Further, the study finds that fundamentalism and religiosity can predict a higher level of ingroup favoritism, and fundamentalism predicts prejudice to outgroups and toward the ones who criticize religious doctrine (for a further study on ingroup favoritism, see, for instance, Fershtman, Gneezy, & Verboven, Reference Fershtman, Gneezy and Verboven2005).
Using a trust game, Burns (Reference Burns2006) identifies a significant pattern of distrust of a Black partner, ‘even by Black proposers’ (Burns, Reference Burns2006: 805), which occurs due to erroneous behavioral stereotypes. White proposers, on the contrary, more likely detract themselves from strategic interaction when paired with a Black partner.
Fershtman and Gneezy (Reference Fershtman and Gneezy2001) study ethnic discrimination in Israeli Jewish society. By using names as a proxy for the affiliation of the Jews to two different Jewish communities, they find that men in particular offer less to especially one of the two communities. Interestingly, the distrust toward the males of this community is perpetuated by the men themselves.
Studies on cross-religion within a country have been conducted, among others, by Johansson-Stenman, Mahmud, and Martinsson (Reference Johansson-Stenman, Mahmud and Martinsson2009, Reference Johansson-Stenman, Mahmud and Martinsson2013) and Chuah, Fahoum, and Hoffmann (Reference Chuah, Fahoum and Hoffmann2013). Exploring trust ascriptions among Hindu and Muslim households in Bangladesh, Johansson-Stenman et al. (Reference Johansson-Stenman, Mahmud and Martinsson2009) find no evidence of social preferences or discrimination among both groups (for an extension to that study, see Johansson-Stenman et al., Reference Johansson-Stenman, Mahmud and Martinsson2013). However, a survey conducted in addition to behavioral experiments reveals that both groups ascribe higher trust to their own respective religious groups. Further, Hindus exhibit lower levels of general trust, but they trust Muslims more than vice versa.
Whether India's societal fractionalization goes hand in hand with distrust between its main religious groups, i.e., Hindus and Muslims, is explored by Chuah et al. (Reference Chuah, Fahoum and Hoffmann2013). In fact, mutually lower offers between Hindus and Muslims tested in Mumbai reveal an intergroup bias and confirm the distrust between the groups. Contrary to the results of Johansson-Stenman et al. (Reference Johansson-Stenman, Mahmud and Martinsson2009), Chua et al. (Reference Chuah, Fahoum and Hoffmann2013) find that Hindus have greater trust, as well as greater expectations of others’ trust.
Noteworthy are studies exploring cross-political systems within a country by focusing on solidarity behavior. Ockenfels and Weimann (Reference Ockenfels and Weimann1999) study solidarity behavior in East and West Germany and found, surprisingly, that Eastern Germans behave more selfishly than West Germans, despite the socialist legacy of East Germany. Interestingly, the experiment was repeated more than a decade later, but the result remained the same. Compared to West Germans, East Germans showed significantly less solidarity (Brosig-Koch, Helbach, Ockenfels, & Weimann, Reference Brosig-Koch, Helbach, Ockenfels and Weimann2011). The authors conclude that political values may change faster than social behavior.
Social identity and discrimination
Since the majority of industrialized countries have become more multicultural during the last two decades, experiencing an ‘increase in ethnic and social heterogeneity’ (Chen & Mendel, Reference Chen and Mengel2016: 1), challenges of cultural heterogeneity within a country become important to understand in more detail. Hence, I have identified a further recent trend of CCEE research that focuses on social identity, which is shaped by culture and can lead to, e.g., discrimination when a person sharply distinguishes between in- and outgroup (for several studies dealing with these questions, see the recently published special issues edited by Chen and Mengel, Reference Chen and Mengel2016).
By focusing on trust and trustworthy behavior, Chuah, Gächter, Hoffmann, and Tan (Reference Chuah, Gächter, Hoffmann and Tan2016) explore discrimination based on religious affiliation. By using test persons from China, Malaysia, and the UK, the authors find that similarity in religiosity and affiliation supports trust between partners, as it is believed that norms of reciprocity take effect. Further, religious test persons ascribe a higher level of trustworthiness to others of the same religious belief.
Chowdhury, Jeon, and Ramalingam (Reference Chowdhury, Jeon and Ramalingam2016) explore the link between social identities and group conflict. Their research design features a setup consisting of two groups: White Caucasians (incl. test persons from North- and West Europe, the US, and Canada), and East Asians (incl. test persons from Japan, South Korea, China, Hong Kong, Macau, and Taiwan). Performing a group contest, in one setup, the racial composition of the group is not revealed, and in another, race information is disclosed. As a result, in the latter treatment, free-riding is observed to decline but conflict effort increases. This was not observed in the treatment in which race information was hidden. The studies presented in this section are summarized in Table 6.
Studies going beyond national culture as a proxy variable
Note: *Studies analyzing intra-country cultural differences
What about Cross-cultural Experimental Management?
Despite the large variety of research methods, the discipline of management applies to study cross-cultural phenomena, ‘experimentation is surprisingly underemphasized’ (Leung & Su, Reference Leung, Su, Punnett and Shenkar2004: 68). This has been also observed in the field of International Business (IB) studies (Leung et al., Reference Leung, Bhagat, Buchan, Erez and Gibson2005; van Witteloostuijn, Reference Van Witteloostuijn2015). However, that does not mean that there are zero cross-cultural experimental studies conducted in the management studies. The few studies that apply cross-cultural experiments most often use country culture as a proxy variable but rarely go deeper (comp. Kirkman, Lowe, & Gibson, Reference Kirkman, Lowe and Gibson2006). Due to the low number of studies, it is currently difficult to clearly separate, categorize, and systematically summarize research in that field in the same fashion as I have done above for the field of CCEE. Nonetheless, by looking above and beyond disciplinary boarders, there are some scattered examples of relevant studies worth mentioning.
In the management studies, analog to the above review of the strategic games used in CCEE, country culture is often used as a proxy variable for studying cultural differences. For instance, Gelfand et al. (Reference Gelfand, Higgins, Nishii, Raver, Dominguez, Murakami, Yamaguchi and Toyama2002), in the frame of a multi-method approach, include a laboratory experiment in order to study fairness perceptions on negotiations in Japan and the United States. This study provides evidence that, in the United States, the self-serving biases of fairness are more pronounced than in Japan where ‘the self is served by focusing on one's negative characteristics to “blend in” and maintain interdependence with others’ (Gelfand et al., Reference Gelfand, Higgins, Nishii, Raver, Dominguez, Murakami, Yamaguchi and Toyama2002: 833). Contrary, in the United States, standing out and being better than others is positively attributed. Consequentially, the results of the study point toward aspects of social context and country-specific internalized ideals that question the conventional view of rationality in negotiations; this is based on accumulated knowledge that has so far been largely based on research conducted by Western scholars.
An approach to studying cultural differences within a multi-ethnic society can be found in the study of Cox, Lobel, and McLeod (Reference Cox, Lobel and McLeod1991) using the example of the United States. In this study, it is assumed that bi-culturals (i.e., Asian-, Black-, and Hispanic Americans) behave more collectivistically and cooperatively than Anglo Americans, who are assumed to be more individualistic and competitive. Test persons were allocated to different groups consisting of an all Anglo group and an ethnically diverse group consisting of the bi-culturals. The results confirm the assumptions made and point toward the importance of the situational context in which decisions are made. Further it points toward the difficulty of clearly classifying multi-ethnic societies along the line of the, e.g., popular individualism-collectivism dichotomy, since situational context or group constellation trigger respective behavior. Chen and Li (Reference Chen and Li2005) study cooperative decision-making behavior of Chinese and Australian test persons by applying two cross-national experiments. It was assumed that the Chinese test persons would behave less cooperatively than the Australians in a mixed-motive business situation and under the absence of formal and informal sanctions systems. Further, it was assumed that Chinese people abroad behave less cooperatively with foreigners than with fellow Chinese, while Australians would not make this distinction. As a result, the experimental data gathered support both assumptions. The study advances our understanding of the conventionally ad hoc assumptions drawn along the lines of dimensional models of culture as the ‘cultural orientations of Australians and Chinese are more complicated than as suggested by previous findings from Hofstede’ (Chen & Li, Reference Chen and Li2005: 630).
In sum, it can be stated that there is a lack of approaches that go beyond the sole treatment of country culture as a proxy variable. Extant studies use country culture and rather broad or horizontal characteristics (e.g., individualism, collectivism, etc., comp. Jung & Avolio, Reference Jung and Avolio1999, or Koch & Koch, Reference Koch and Koch2007) instead of verticalizing their approach and diving deeper into indigenous antecedents of a culture. Nevertheless, it is a positive development that past research directs attention to, e.g., social context and internalized ideals. However, so far, most research cannot clearly answer of what cultural tiles the social context is comprised and what ideals are in detail and whether and how they in fact influence behavior.
LESSONS LEARNED AND THE MISSING LINK
Echoing earlier implications of research that reflect the progressing decay of the behavioral assumptions of the homo oeconomicus (e.g., Boone, De Brabander, & Van Witteloostuijn, Reference Boone, De Brabander and Van Witteloostuijn1999; Falk & Fischbacher, Reference Falk and Fischbacher2006; Henrich, Reference Henrich2000; Henrich et al., Reference Henrich, Boyd, Bowles, Camerer, Fehr, Gintis and McElreath2001; Kahneman et al., Reference Kahneman, Knetsch and Thaler1986; Thaler, Reference Thaler1988), Stigliz noted in his Nobel prize speech that ‘the economists’ traditional model of the individual is too narrow’ (Stiglitz, Reference Stiglitz2002: 488). While traditional economists have not given the influence of culture on decision-making behavior much thought (Jones, Reference Jones2006), the implications derived from cross-cultural economic experiments are particularly interesting. They provide empirical evidence that the standard assumption of economic behavior (i.e., the homo oeconomicus) would need revision as culture de facto moderates economic behavior. Moreover, a consciousness of fairness in human transactions, a feature the homo oeconomicus does not care about, differs across countries, often in a way so that decisions even cause economic loses. The results from the experiments presented above grant deeper insights into how people from different cultures behave when urged to make critical decisions. If fair behavior and the norms of cooperative behavior deviate based on the prevailing social norms in a community, they are enforced with a different intensity (e.g., some studies found differences in the level of free-riding). Japanese tend to sanction free-riders more strongly than Americans do and Russians punish overly egoistic as well as overly altruistic behavior much more strongly than the Swiss. Generalizations derived from these studies provide useful indications for further inquiry, for instance, how culture influences group decision making. It makes a difference whether a group consists of culturally homogeneous participants or whether decisions are made in heterogeneous groups. Here, some national cultures are affected by the group constellation, whereas others are not. Moreover, religion, an important cultural factor of influence, can also explain differences in decision-making behavior (Chen & Tang, Reference Chen and Tang2009).
Implications based on these results, however, have to be treated with caution. Given the absence of consistently applied standardized effect measures, past CCEE studies have rarely discussed size effects or negative or null results. Analog as to the challenges in conducting a meta-analysis in the psychology discipline (comp. Morris & DeShon, Reference Morris and DeShon2002), CCEE studies often fail to provide enough information to sufficiently assess size effects or use different experimental designs to explore the same phenomena. Replicability gets complicated when different experimental designs estimate different population parameters (Glass, McGaw, & Smith, Reference Glass, McGaw and Smith1981; Morris & DeShon, Reference Morris and DeShon1997; Ray & Shadish, Reference Ray and Shadish1996). Spending more attention on interpreting size effects as well as discussing negative or null results, would increase the quality of insights based on the results generated (Lewin et al., Reference Lewin, Chiu, Fey, Levine, McDermott, Murmann and Tsang2016).
However, despite all its achievements, CCEE has not yet exhausted all of its potential. Thought progress has been made since the 1990s when differences in decision-making behavior of test persons from different countries was assumed to be caused simply by cultural differences, to more recently when questions of behavioral differences were explored through a focus on ethnicity, individual values, or social identity, still one may argue that CCEE could dive deeper into questions that make us better understand the effect of culture on decision-making behavior. For instance, Oosterbeek et al. (Reference Oosterbeek, Sloof and Van De Kuilen2004) is missing a more detailed understanding about the ‘underlying factors’ of behavioral differences. Also CCEE needs ‘to identify the components of particular cultural systems and the respective effects these have on economic behavior’ (Chuah et al., Reference Chuah, Hoffmann, Jones and Williams2007: 46). So how can indigenous management scholars bring in their competencies? The following section proposes how research can be designed in the future to integrate a higher degree of contextualization in order to study the influence of cultural factors on decision-making behavior in more depth.
TOWARD EXPERIMENTAL INDIGENOUS MANAGEMENT RESEARCH
Suggestions for Future Research
Just as the economic rise of many emerging countries has generated higher demand for explaining the reasons of their success (e.g., their management practices), awareness has grown among the scholarly community that conventional management theories, coined mostly by Western scholars’ ideas, are not universal per se, or they are unable to fully explain the managerial principles of many emerging market economies (Van de Ven & Jing, Reference Van de Ven and Jing2012; Werner, Reference Werner2002; Zhao & Jiang, Reference Zhao and Jiang2009). The currently flourishing field of Indigenous Management Research (IMR) aims to better understand distinctive cultural factors and their influence on management (Leung, Reference Leung2012; Li, Leung, Chen, & Luo, Reference Li, Leung, Chen and Luo2012; Li, Reference Li2015; Tsui, Reference Tsui2007; Van de Ven & Jing, Reference Van de Ven and Jing2012). By taking the cultural context explicitly into account, IMR aims to overcome the scientific paradigms developed predominately by Western scholars in the past (Banerjee & Prasad, Reference Banerjee and Prasad2008; Barney & Zhang, Reference Barney and Zhang2009; Fang, Reference Fang2009; Lin, Reference Lin2002; Tsui, Reference Tsui2004). Following the call for high-quality IMR (Tsui, Reference Tsui2004), this paper argues that an experimental methodology can make a contribution to IMR by drawing on the strengths of both IMR (i.e., contextualization) and CCEE (i.e., methodology). Typically, IMR uses comparative case analysis, participant observations, interviews, questionnaire-based surveys, or ethnographic inquiry (Fang, Reference Fang2012; Ma, Reference Ma2012; Van de Ven & Jing, Reference Van de Ven and Jing2012; Wu, Huang, Li, & Liu, Reference Wu, Huang, Li and Liu2012). Since experiments have not been popular in IMR as of yet, how can they help? Experimental approaches would well suit to drive, for instance, guanxi research to a next level (Chen, Chen, & Huang, Reference Chen, Chen and Huang2013; Li et al., Reference Li, Leung, Chen and Luo2012; Song, Cadsby, & Bi, Reference Song, Cadsby and Bi2012). So far the bulk of research on guanxi has been either conceptual (Fan, Reference Fan2002, Reference Fan2007; Luo, Reference Luo2008) or based on interviews (Anderson & Lee, Reference Anderson and Lee2008; Zolkiewski & Feng, Reference Zolkiewski and Feng2012) or questionnaires (Yen, Barnes, & Wang, Reference Yen, Barnes and Wang2011; Yi & Ellis, Reference Yi and Ellis2000). An experimental research design would contribute to current knowledge by exploring how subjects behave in a close-to-reality environment as members of a guanxi network. Whereas currently, guanxi is often treated as an en bloc phenomenon, or an aggregate that actually consists of several antecedents, approaches are needed that test the effects of particular antecedents of the guanxi construct. Therewith, quantitative data could be generated that would add to the discourse of whether guanxi is still influential today or whether it is transforming or declining (Brennan & Wilson, Reference Brennan and Wilson2010; Guthrie, Reference Guthrie1998; Lee & Dawes, Reference Lee and Dawes2005).
By using the case of Korea, the study presented in the next section exemplifies an approach for how indigenous cultural context variables can be extracted, integrated, and tested in the frame of an experimental indigenous research design that leads to a better understanding of how and in what way they engender influence on actual behavior.
Example – Testing cultural context variables (CCV)
A showcase example of an experimental indigenous research design is the study on decision-making behavior and cultural context variables by Horak (Reference Horak2016). In order to understand the influence of distinctive cultural variables of the Korean cultural context on decision-making behavior, this study integrates high-impact cultural context variables into the research design. Thus, it responds to the call for a research design that leads to a better understanding of the underlying factors that influence decision making. In this study, the ‘primary objective is to investigate how cultural values embodied in the Korean cultural context influence the decision-making behavior (…)’ (Horak, Reference Horak2016: 282). The study features a two-stage design consisting of expert interviews in the first stage and behavioral field experiments in the second. The first stage serves to identify important variables of the cultural environment that influences decision making in Korea. The results of the interviews showed that a variety of factors play a role, with, among the most influential ones being seniority, as a crucial variable derived from the moral philosophy of Confucianism, which is pervasive throughout Korean society. Further, the interview partner reported that education-based ties (university and/or high school), regional origin, and family ties play an important role as these factors represent central elements of the social identity of Koreans. Going deeper, the interviews revealed that these three ties are commonly known as so-called yongo ties. The first syllable of this Korean term, ‘yon’, translates into ‘tie’ and the second syllabi ‘go’ carries the meaning that the tie exists for a reason, and the reason is usually a shared background composed of the same home town (i.e., regional origin, in Korean: jiyon [지연]), attendance at the same high school and/or university (in Korean: hakyon [학연]), and blood ties (i.e., family ties, in Korean: hyulyon [혈연]). It can be stated that Korean society, especially the Korean business and political sphere, consists of pervasive informal ties based on exactly these yongo ties. Yongo ties usually last for life and their strengths depend on how intensively they are maintained. Yongo is characterized as immutable, and to some extent, predefined (i.e., given by birth) with education-based ties today being the exception, as they can be freely chosen.
As a result of the interviews, a set of selected variables, the cultural context variables (CCV), were, in the second stage, integrated in an ultimatum game that was played in two variants. One variant featured the CCVs and the other, for benchmark reasons, the standard version of the ultimatum game. In the frame of a paper-and-pencil experiment, the CCVs were communicated in writing to the respective decision maker. By following the same protocol, these two games were played in Korea among a Korean subject pool as well as in Germany among a German subject pool. The latter represented a culturally different environment that was chosen for comparative reasons. It was assumed that if the CCVs represent distinctive Korean variables of the cultural environment they would influence the decision-making behavior of the Korean test persons but would not deploy influence among a German subject pool. Interestingly, these assumptions could be confirmed. In more detail, by distinguishing decision-making behavior among the sexes, it turned out that male Koreans very strongly responded to the CCVs. Depending on their randomly assigned position in the game, they either behaved egoistically in relation to their offer behavior or responded by not accepting even an advantageous offer. On the contrary, Korean females did not change their behavior independent of the game condition and respective randomly assigned role. Interestingly, a change of behavior could not be observed among the female and male test persons in Germany, where the CCVs had no influence.
The research approach taken in this study goes beyond the conventional country culture comparison and individualism-collectivism ad hoc interpretation by first identifying and extracting influential antecedents of the cultural context that may influence behavior. This approach represents a small step toward a better understanding of the components of a cultural system and opens up new avenues for future research by putting the identification and testing of specific high-impact CCVs of a respective cultural environment at the center of the approach.
Prerequisites, Challenges, and a Process Model for Experimental Indigenous Management Research
In order to conduct such research, a precondition is the possession of profound knowledge about the cultural environment and its specific context factors. Experimenters unfamiliar with the prevailing behavioral ideals of respective cultures need to be aware that a lack of knowledge of indigenous culture runs the risk of testing behavioral dimensions that are influenced by prevailing behavioral norms without knowing them. As a result, data may easily be misinterpreted. Close research cooperation with local scholars familiar with the environment hedges this risk (Tsui, Reference Tsui2007; Van de Ven & Jing, Reference Van de Ven and Jing2012; Zhao & Jiang, Reference Zhao and Jiang2009).
Alternatively, once behavioral differences are indentified across cultural lines, applying a multi-method design may help, for example, by conducting contextualized qualitative research using ethnographic research methods or narrative interviews in order to identify high-impact context variables. In a further step, the identified context variable(s) could be integrated into an experimental research design and tested in the field, as proposed in the study by Horak (Reference Horak2016) presented above (Figure 1).
Proposal of a process model for experimental indigenous management research
In order to increase validity, a triangulation of data could be conducted by including a third stage after the experiment: a post-experimental questionnaire or a post-experimental interview (e.g., Greiner, Caravella, & Roth, Reference Greiner, Caravella and Roth2014; Lönnqvist, Verkasalo, Walkowitz, & Wichardt, Reference Lönnqvist, Verkasalo, Walkowitz and Wichardt2015; Maciejovsky, Kirchler, & Schwarzenberger, Reference Maciejovsky, Kirchler and Schwarzenberger2007). Both instruments serve to gain a better understanding of the motives and inferences of particular decisions by the test persons. In terms of indigenous research on East Asia, for instance, when certain antecedents of seniority are explored, the filial piety scale (Ho, Reference Ho and Bond1996; Ho & Lee, Reference Ho and Lee1974) could be administered after the experiment. Depending on the nature of the research subject, the experimentor needs to decide whether a post-experimental questionnaire or a post-experimental interview suits better. This decision is, in principle, similar as the decision made before the experiments when the appropriate (pre-) interview instrument needs to be selected, for instance, narrative, semi-structured, or structured. While the nature of the research subject as well as the contextual knowledge level of the research will determine the decision, it may be suggested that a very novel theme and a rather low level of indigenous context knowledge would rather point toward the selection of narrative interviews, whereas for follow-up studies that are building on an existing body of knowledge, a questionnaire-based post-experimental survey may be suitable. It remains an individual case-by-case decision. Taken together, these suggestions form a process model of experimental indigenous management research consisting of three stages, starting with the identification of variables of the cultural context (stage 1) and then proceed with applications to be aimed at testing and understanding the effects of the respective variables on behavior (stage 2 and 3). The proposed process model is shown in Figure 1.
The approach proposed would enable the determination in greater detail of the circumstances under which culture is and is not influential. The resulting findings can contribute to existing knowledge, for example, in the area of cooperation behavior in individualistic and collectivistic culture groups. It can also contribute to a better understanding of what determines the establishment of trust and trustworthiness across and between cultures, how fairness is interpreted across cultures, and how it is enforced. These questions can be answered by integrating context-specific variables into the research design.
Another approach for achieving progress in better understanding culture is to investigate deeper the enduring paradigm of cultural dimensions. Experimental economic approaches can contribute to this field. In the past, concerns have been raised that conventional cultural categories (Hofstede, Reference Hofstede1980, Reference Hofstede1991) are not sufficiently suitable for explaining national cultures in a satisfactory way, and that they treat national cultures as static constructs (Fang, Reference Fang2012; Van de Ven & Jing, Reference Van de Ven and Jing2012). For instance, Japanese society is generally viewed as highly collectivistic (Hofstede, Reference Hofstede1991) in which collective ingroup norms prevail and guide individual behavior. However, empirical research from the discipline of social psychology suggests that this claim should be treated with caution. Most Japanese people regard themselves as individualistic rather than collectivistic, but think that other Japanese people internalize collectivistic norms (Hirai, Reference Hirai2000). If collectivistic behavior helps secure resources, incentives are provided to act in such a manner. In that case, collective behavior is rather strategic, but is not based on internalized norms of collectivism as the conventional cross-cultural management literature indicates. Yamagishi, Hashimoto, and Schug (Reference Yamagishi, Hashimoto and Schug2008) suggest that the nature of Japanese collectivism has evolved through ‘pluralistic ignorance’. Empirical evidence for this argument is provided by Ohbuchi and Saito (Reference Ohbuchi and Saito2007), who find that the maintenance of conflict avoidance behavior in Japan is a self-fulfilling prophesy. Other studies provide empirical evidence that Japanese collectivism should in fact be regarded as an individualistic strategy rather than an internalized, norm-based one (Yamagishi, Jin, & Kiyonari, Reference Yamagishi, Jin and Kiyonari1999; Yamagishi, Terai, Kiyonari, Mifune, & Kanazawa, Reference Yamagishi, Terai, Kiyonari, Mifune and Kanazawa2007). These are important findings toward a more grounded understanding of culture based on a deeply contextualized experimental approach.
IMR and CCEE – Meeting in the Middle
Economic experiments identify the behavioral differences derived from cultural differences but seldom provide an explanation based on cultural heritage in the form of the indigenous independent variables included in the research design. As a result, various studies lack detail and background concerning the factors from which cultural differences can be derived. On the other hand, IMR seldom uses experimentation in order to test independent variables typical for the country being studied. There is a great possibility that both disciplines may ‘meet in the middle’.
Leung et al. (Reference Leung, Bhagat, Buchan, Erez and Gibson2005) see the advantages of the experimental methodology, especially in its ability to examine the causality of individual and group behavior (see also van Witteloostuijn, Reference Van Witteloostuijn2015). Therefore, it is possible to determine in greater detail the circumstances under which culture is and is not influential. The resulting findings can contribute to the advancement of management theory, for example, in the area of cooperation behavior in individualistic and collectivistic culture groups. It can also contribute to a better understanding of what determines the establishment of trust and trustworthiness across and between cultures, and how fairness is interpreted across cultures and how it is enforced. These questions can be answered by integrating context-specific variables into the research design. This proposal is in line with the recent call in economics to better understand the specific contextual variables that explain culturally-induced differences in decision-making behavior (Chuah et al., Reference Chuah, Hoffmann, Jones and Williams2009; Oosterbeek et al., Reference Oosterbeek, Sloof and Van De Kuilen2004). A key strength of economic experiments compared with other methodologies is that they allow researchers to observe closer to real-world behavior or ‘attitudes in action’ since they are incentivized by monetary benefits in a controlled environment (e.g., in an experimental lab) in which influencing factors can be controlled.
The advantage that experiments can be designed in a way so that they can be replicated by other scholars contributes to attempts to replicate findings, which again adds to a better understanding about the consistency of the findings as well as size effects. Overall, I note that although many studies discuss null and negative findings at times comprehensively, studies discussing effect sizes explicitly and in detail are rare. This can possibly be explained by the absence of a common standard in the field of CCEE, which is especially the case for qualitative studies using content analysis (comp. Onwuegbuzie, Reference Onwuegbuzie2003). The interpretation of different levels of variance would be especially important for an experimental approach to IMR in order to gain better insights into the meaningfulness of a statistical effect. In more detail, the interpretation of the effect size is important in order to understand better how strong, for instance, a cultural variable (independent variable) affects the dependent variable. Depending on the case, even small effects shouldn't be automatically labeled and interpreted as weak or trivial per se, since they can still represent impressive support for an examined phenomenon (Cortina & Landis, Reference Cortina, Landis, Lance and Vandenberg2009). In addition, in order to understand better the practical relevance of its inquiry (see also Aguinis, Werner, Abbott, Angert, Park, & Kohlhausen, Reference Aguinis, Werner, Abbott, Angert, Park and Kohlhausen2010), future research designs can apply complementary pre- or post-experimental interviews in order to better access practical implications of the respective themes explored.
Subject Pool Selection and Incentive Compatibility
What the right subject pool may be for what particular research question has been debated in the field of experimental economics (e.g., Alevy, Haigh, & List, Reference Alevy, Haigh and List2009; Croson & Donohue, Reference Croson and Donohue2006) as well as in International Business studies (e.g., Bello, Leung, Radebaugh, Tung, & Van Witteloostuijn, Reference Bello, Leung, Radebaugh, Tung and Van Witteloostuijn2009). While it appears to remain essentially a case-by-case decision, in general, it can be recommended that the selection of an appropriate subject pool should be aligned with the respective research question. Whereas for very basic oriented research that aims to explore fundamental patterns of human behavior, a subject pool consisting of – next to adults – young test persons (i.e., students) may be a valid option. However, by drawing on Sears (Reference Sears1986), Bello et al. (Reference Bello, Leung, Radebaugh, Tung and Van Witteloostuijn2009) recommend not using undergraduate students in particular, as they may likely have ‘less crystallized attitudes, less-formulated senses of self, stronger cognitive skills, stronger tendencies to comply with authority, and more unstable peer group relationships’ (Sears, Reference Sears1986: 515). More clearly, when it comes to the study of proximate issues that focus more on a specific (practical) context, students with no real-world managerial experience would have difficulties responding to strategic decisions in a multifaceted international context. For this case, there is hardly a viable alternative to a subject pool of managers. Executive MBA students ‘who are involved in high-level decision making could be appropriate, however’ (Bello et al., Reference Bello, Leung, Radebaugh, Tung and Van Witteloostuijn2009: 362).
Questions of subject pool compositions lead to questions of incentive compatibility. Finding a suitable level of incentives for a demographically similar group is crucial, as large deviations (i.e., overpay as well as underpay) may bias decision-making behavior. A suitable level of reward for student subjects is generally seen in the hourly compensation rate that a university would pay for a student assistant (Hennig-Schmidt et al., Reference Hennig-Schmidt, Li and Yang2008). Rewarding management includes at least three major challenges. Gaining access to a larger group of this naturally busy type of test persons is one of them. Further, while it can be debated whether monetary incentives represent a suitable instrument to motivate a generally well-paid group of people, finding a suitable level of rewards is difficult, and as that level will very likely be much higher compared to that of students, larger research budgets are required, which may be a burden to acquire for the average scholar. As these potential constraints need to be taken into account before preparing a research project, there are, however, examples where scholars successfully solved the issue of budget constraints and gained access to a large number of CEOs willing to participate in a behavioral experiment, such as the study by Opper, Nee, and Holm (Reference Opper, Nee and Holm2017). In a recent AMJ (Academy of Management Journal) article, using a behavioral experiment, Opper et al., (Reference Opper, Nee and Holm2017) study risk aversion of Chinese CEOs by focusing on the association between risk aversion and the reliance on social relationships. The subject pool consists of 345 CEOs of private manufacturing companies. Whereas on average the CEOs earned USD 47 each (in local currency: CNY 289), the maximum they were able to earn in the experiment was ca. USD 95 (corresponding to CNY 580 in local currency). In terms of the suitability of the level of the reward for CEOs in China, the authors believe that ‘given the modest amount of time needed for completion [of the experiments] – [the reward] presented a credible incentive even for CEOs’ (Opper et al., Reference Opper, Nee and Holm2017: 1512).
In sum, scholars need to think critically about what subject pool may best suit the respective research questions. Once fundamental behavioral phenomena are explored, next to adults, graduate or MBA students might be a suitable option (Bello et al., Reference Bello, Leung, Radebaugh, Tung and Van Witteloostuijn2009). In case questions relevant to a business setting are explored that target clearly a managerial context and aim at generalizing managerial behavior, undergraduate students very likely represent an unsuitable group of test persons, and managers would need to be recruited. Executive MBA students, as recommended by Bello et al. (Reference Bello, Leung, Radebaugh, Tung and Van Witteloostuijn2009), depending on the research context, may represent an alternative.
CONCLUSION
As several cross-cultural economic experiments (CCEEs) provide evidence that culture may moderate economic behavior, explaining why behavior differs has so far remained a challenge. Recently, the need to better understand the distinct nature (or underlying factors) that cause culturally-induced behavioral differences has been raised in experimental economics (Chuah et al., Reference Chuah, Hoffmann, Jones and Williams2007; Oosterbeek et al., Reference Oosterbeek, Sloof and Van De Kuilen2004). Whereas CCEE is, methodology-wise, fixed to behavioral experiments, indigenous management research (IMR) includes a large variety of qualitative and quantitative research methods, except experiments. Following the call by Tsui (Reference Tsui2004) for advancing indigenous research, in this paper I argue that an experimental methodology can advance IMR by combining the strengths of both approaches. In simpler terms, it is suggested that by pursuing an experimental indigenous research approach, distinctive cultural characteristics can be extracted in a first step, and then in a second step integrated into a controlled behavioral experiment in order to test their effects on behavior. A third stage can be implemented by conducting either post-experimental interviews or using a post-experimental questionnaire in order to support validity.
APPENDIX I
Decision Options and Corresponding Results in a Standard Ultimatum Game (UG)
APPENDIX II
Decision options and corresponding results in a standard public good game (PG)
APPENDIX III
Decision Options and Corresponding Results in a Standard Trust Game (TG)
Appendix IV
Third-party Punishment Game (3PPG) - Payoffs of Three Player (a) with Punishment and (b) without Punishment (Wxample)
APPENDIX V
Other Common Game Applications
