An experimental study on the relevance and scope of nationality as a coordination device.
Stoddard, Olga ; Leibbrandt, Andreas
I. INTRODUCTION
The beginning of the new millennium has been marked by increased
social, economic, technological, and cultural integration. These
phenomena, in turn, promote contact between individuals from different
countries and cultural backgrounds putting higher demands on the
coordination of actions. (1) In this context, it is important to
understand if and how coordination depends on the cultural backgrounds
of the participating actors. One stylized fact in the economic
literature on coordination is that in the absence of additional
information, people tend to use a solution that seems relevant, natural,
or special to them, known as a focal point (Schelling 1960; Crawford and
Haller 1990; Mehta et al. 1994; Sugden 1995; Camerer 2003). Nationality
may be such a focal point and hence one may hypothesize that nationality
serves as a coordination device and that coordination is less likely to
occur if interactions take place between partners from different
nationalities (Schelling 1960; Sugden 1986).
The existing body of literature provides grounds for such a
hypothesis. Efferson et al. (2008), for example, suggest that
coordination among compatriots may be facilitated in strategic settings
with multiple equilibria because of in-group favoritism, particularly
when trivial groups evolve into cultural groups. Theoretically, Chen and
Chen (2011) use a group-contingent social preference model to show that
an induced salient group identity can lead to higher in-group
coordination on the efficient high-effort equilibrium in the
minimum-effort coordination games. As nationality represents an
important dimension of individual identity, we expect nationality to
matter in coordination settings and conjecture that individuals
coordinate better when they interact with their compatriots as compared
to non-compatriots or as compared to subjects whose nationality they do
not know.
In this study, we use an experimental approach to investigate
whether and under which conditions nationality serves as a coordination
device in coordination games with multiple equilibria (Crawford and
Haller 1990; Cooper et al. 1990; Van Huyck et al. 1990; Crawford 1995).
We made use of the diverse student body at the University of Hawaii and
recruited an equal amount of Japanese, Korean, and Chinese nationals.
The subjects play simple stag-hunt coordination games with a
payoff-dominant equilibrium (PDE) and a risk-dominant equilibrium (RDE),
and interact both with their compatriots and with participants from the
other two countries. To test for the robustness of the relevance of
nationality as a focal point, we provide subjects with different levels
of information about their partner; (1) no information, (2) information
only about their partner's nationality (i.e., nationality is
salient), and (3) information about their partner's nationality and
some other presumably irrelevant characteristics such as hair color
(i.e., we make nationality less salient). In addition, we observe the
subjects' level of pro-sociality toward subjects from other
nationalities and compatriots in ultimatum and dictator games to
investigate whether potential differences in coordination are driven by
in-group favoritism (Tajfel and Turner 1979).
Our findings show that nationality can indeed serve as a
coordination device. Subjects are more likely to try to coordinate on
the PDE if common nationality is salient as compared to if subjects do
not know the nationality of their partner. However, if nationality is
non-salient, nationality does not serve as a coordination device, that
is, subjects are not more likely to try to coordinate on the PDE if
their partner has the same nationality. Moreover, we do not find that
subjects are less likely to try to coordinate on the PDE if their
partner has a different nationality as compared to when the
partner's nationality is unknown.
To the best of our knowledge, this is the first experimental study
analyzing inter-cultural coordination between different nationalities in
a systematic manner and inter-cultural behavior where the salience of
nationality is experimentally manipulated. Probably the closest related
study is that of Brandts and Cooper (2007) who compare the behavior of
subjects in the United States and Spain and observe higher levels of
coordination in the United States. This study, however, does not
investigate coordination between subjects from different nationalities.
Related to our findings is also the study by Crawford et al. (2008) in
which the authors find that miscoordination (MIS) increases if the
salience of focal points is reduced by minimally changing payoff
constellations, and Holm (2000) who finds that information about gender
affects coordination in a battle-of-sexes game.
In addition, a closely related literature in social psychology
studies economic behavior in the context of social identity theory and
nationality in particular. Examples include Kuwabara et al. (2007) and
Takshashi et al. (2008). Both studies report the results of web-based
laboratory experiments where individuals from two or three countries
play a variant of the trust game online under two experimental
conditions--one in which the partner's nationality is known, and
another in which the partners remain anonymous. Similarly, Yamagishi et
al. (2005) had participants from Japan and Australia play five rounds of
Prisoner's Dilemma game, each time with a different partner,
varying the extent of mutual knowledge of the partners regarding each
other's nationality. However, while these studies all use
nationality in their experimental conditions, they do not experimentally
manipulate the salience of nationality, as we do in our study.
Our study contributes not only to the literature on coordination
games which are characterized by the existence of multiple Nash
equilibria (Crawford and Haller 1990; Cooper et al. 1990; Van Huyck et
al. 1990; Crawford 1995) but also to the literature comparing behavior
across cultures (Roth et al. 1991; Okada and Riedl 1999; Anderson et al.
2000; Henrich 2000; Henrich et al. 2001; Brandts et al. 2004), between
cultures (Fershtman and Gneezy 2001; Buchan et al. 2006; Chuah et al.
2007; Bornhorst et al. 2008; Kuwabara et al. 2007; Yamagishi et al.
2008), and more generally to the literature on in- and out-group
behavior (Sherif et al. 1961; Tajfel and Turner 1979; Kollock 1998;
Eckel and Grossman 2005; Goette et al. 2006; Takshashi et al. 2008;
Charness and Rustichini 2007; McLeish and Oxoby 2007; Chen and Li 2009)
which typically reports that individuals treat in-group members better
than out-group members.
Most of these studies do not manipulate the salience of group
membership. Exceptions are Eckel and Grossman (2005) and Charness and
Rustichini (2007), which create minimal groups to study cooperation and
increase group identity by team goal attainment or passive audiences.
Consistent with our findings, these two studies point out that the
saliency of group membership affects behavior. In contrast to these
studies, we investigate coordination among natural groups of people with
different nationalities and decrease group identity by providing
subjects with different levels of information about their partners.
II. HYPOTHESES
In this section, we present our research hypotheses regarding
subject behavior in coordination games with multiple equilibria as
related to nationality, as motivated by the theoretical and empirical
findings which we discuss in the following section.
HYPOTHESIS 1: Nationality serves as a coordination device in
coordination games with multiple equilibria.
This hypothesis implies that relative to a condition where players
have no information about their partner's nationality, information
about common nationality serves as a focal point and enhances
coordination between players. Theoretically, this hypothesis is
motivated by a group-contingent social preference model (Chen and Chen
2011) that shows that an induced salient group identity can lead to
higher in-group coordination on the efficient high-effort equilibrium in
the minimum effort coordination games. We extend their analysis to
natural group identity.
Furthermore, in order to explore not only if but also under what
conditions nationality serves as a coordination device, we hypothesize
that varying the salience of information about nationality may have an
impact on the effectiveness of nationality as a coordination device.
Empirical findings of Eckel and Grossman (2005) and Charness and
Rustichini (2007) point to the importance of manipulating the salience
of group membership in minimal groups, and we expect that it is even
more important in natural groups.
HYPOTHESIS 1a: Nationality serves as a coordination device if
nationality is salient.
HYPOTHESIS 1b: Nationality serves as a coordination device if
nationality is non-salient.
These hypotheses imply that relative to having no information about
the partner's nationality, having either salient or non-salient
information about common nationality increases coordination between
players.
HYPOTHESIS 2: Coordination is more difficult if subjects know that
their partner has a different nationality.
This hypothesis is motivated by Tajfel and Turner's (1979)
theory of social identity and an extensive empirical literature on in-
and out-group behavior (cited in the previous section), which generally
finds that subjects treat in-group members better than the out-group. We
therefore expect coordination to be less likely when subjects know they
are matched with a non-compatriot relative to when they are matched with
a compatriot.
Together, hypotheses la, lb, and 2 form the basis of our
experimental design, which we present in the next section.
III. EXPERIMENTAL DESIGN
Our experiment consists of four parts: (1) a short pre-experimental
questionnaire to collect the subjects' demographic information,
including nationality; (2) instructions and control questions for the
games (the complete set of experimental instructions and questionnaires
is given in the Appendix); (3) the games; and (4) a short
post-experimental questionnaire before subjects are paid privately.
All subjects first took part in three ultimatum games and three
dictator games before they played three coordination games. Each game
was presented to the subjects separately and they received no
information about the behavior of other subjects or the outcomes of the
games until the end of the experiment. The roles remained the same
throughout the experiment, that is, a subject who was assigned a role of
a proposer in the ultimatum game, was also a proposer in the dictator
game. Players were randomly paid for one of the three ultimatum games,
one of the three dictator games, and one of the three coordination
games. In addition, we used the perfect stranger matching, so no player
knew the identity of his/her co-player and no player was ever matched
with the same player twice. Therefore, there is no theoretical reason to
believe that the behavior in the coordination games is contaminated by
the preceding games. (2) By taking into consideration the behavior in
the ultimatum and dictator game, we can control for the possibility that
the treatment differences in the coordination game are driven by some
type of in-group favoritism. (3)
The three coordination games (as well as the three ultimatum and
dictator games) differ according to the matching and were presented to
participants in random order. In one decision, a subject was matched
with a person from the same nationality, and in the other two he/she was
matched with subjects from other nationalities than his/her own. The
matching algorithm is presented in Figure 1. The experiment was
programmed using Z-tree software (Fischbacher 2007).
There are three between-subject treatments in this experiment which
differ by the amount of information subjects have about their matched
partners: control (treatment C), salient information about
partner's nationality (treatment S), and non-salient information
about partner's nationality (treatment NS). In C, a subject
receives no information about his/her partner. In S, the only piece of
information a subject receives is the nationality of his/her partner. In
NS, a subject receives information about the age, university status
(freshman, sophomore, junior, senior, or graduate student), eye color,
hair color, and nationality of his/her partner. No subject participated
in more than one treatment.
The coordination game we study is a symmetric two-person stag-hunt
game in which the subjects simultaneously choose either A or B. The
earnings are determined depending on their choice and the choice of
their match according to the payoff matrix presented in Appendix B. For
example, if Participant 1 chooses A, then Participant 2 earns seven
tokens if he/she chooses A and one token if he/she chooses B. Each token
is worth $1 if the game is chosen for payment. Note that choice B is
riskier for the subject as he/she could either make one or nine tokens
depending on the choice of his/her match. By contrast, choice A is less
risky as it secures at least seven tokens regardless of the other
subject's decision. There are two Nash equilibria in this game:
(A,A) which is the RDE, and (B,B) which is the PDE. Chamess (2000)
studies this game with the same parameter specifications.
In the ultimatum games, a proposer has to decide how to divide ten
tokens between him/her and a responder. If the responder accepts, the
offer is implemented; however, if he/she rejects both player types
receive zero tokens. We implemented the strategy method for the
responders, that is, the responder had to determine his/her minimal
acceptable offer before he/she knew the actual offer of the proposer. In
the dictator games, the proposer has also to decide how to divide ten
tokens between him/her and a responder but the responder cannot reject
the offer.
A total of seven sessions were conducted in the UH Experimental
Laboratory in April 2009 (two sessions for treatments C and NS each,
three sessions for treatment S). (4) Sessions typically included 18
participants: 6 Japanese, 6 Korean, and 6 Chinese students. A total of
126 subjects participated in the experiment (42 Japanese, 38 Korean, 42
Chinese, and 4 other nationalities). (5) These nationalities were chosen
because of their significant representation at the University of Hawaii
as well as their close regional ties. Choosing participants from the
same region allows reducing possible cross-regional factors that may
affect subjects' behavior and potentially contaminate the results
of the experiment. One challenge is that these participants may be less
representative for the population residing in the respective countries
and consist of those who only weakly identify with their nationality.
Thus, we have made every possible effort to recruit subjects who still
have strong ties to their nationalities and our sample consists of
mainly students who have not been abroad for a longer period (55% of our
subjects had been abroad for less than 2 years) and still identify
strongly with their nationality (98.3% of the subjects reported to
strongly identify with their nationality).
Our subject pool has the following characteristics. Sixty-eight
percent of the subjects are female; 57% identify themselves very
strongly with their nationality, 41% somewhat strongly, and 2% not at
all. Fifty-five percent of the subjects report to have a grade point
average (GPA) of 3 (=mean grade in their university classes, out of 4),
and 37% a GPA of 4. With regard to the information subjects receive
about their partner in the NS treatment, we observe that 55% have black
and 44% brown eyes, 86% have black and 13% brown hair. The mean age is
25 years; 38% are graduate, 25% are senior, 21% junior, and 16%
sophomore and freshman students. Each subject participated in only one
session. Sessions lasted approximately 45 minutes and the subjects
earned on average $13 ($5 show-up fee, plus their earnings from the
experimental sessions). Table 1 presents a summary of the sessions.
IV. RESULTS
We start this section with a brief overview of behavior in the
coordination, dictator, and ultimatum games before we investigate
whether nationality serves as a coordination device. In the coordination
games, we find that overall the risky alternative B was chosen 44.1 % of
the time (156 out of 354). There are no significant differences in the
mean probability to choose B among nationalities. Japanese choose B with
a probability of .467, which is similar to Chinese (.443; Fisher's
exact test, two-sided) and Koreans (.409; Fisher's exact test, p=
.427, two-sided). We observe that 18.1% of the pairings achieve the PDE,
29.9% the RDE, and 52% miscoordinate.
Table 2 summarizes the means of all three treatments under the two
different matchings (same or different nationalities) for the
coordination, ultimatum, and dictator games. In the control treatment,
we observe that 42.6% of the subjects choose B in the coordination game
(Japanese: 47.2%, Chinese: 41.7%, and Korean: 38.9%). In the salient
treatment, 45.8% choose B (Japanese: 47.1%, Chinese: 41.2%, and Korean:
50%), and in the non-salient treatment 43.1% (Japanese: 45.7%, Chinese:
51.4%, and Korean: 31.5%).
In the dictator game, the mean token amount sent is 3.26 (treatment
C = 3.33, S = 3.00, and NS = 3.55). In the ultimatum game, the mean
token amount sent is 4.75 out of 10 (treatment C = 4.62, S = 5.12, and
NS = 4.33) and the mean minimum acceptable offer is 3.12 (treatment C =
3.13, S = 3.35, and NS = 2.78). The mean individual behavior in the
three coordination games is not significantly correlated to the mean
individual behavior in the three dictator games (r = .059, p = .654),
the mean individual proposer behavior in the ultimatum games (r = .194,
p = .138) or the mean individual minimal acceptable offer in the
ultimatum games (r = .073, p = .574).
We now address the question of whether nationality serves as a
coordination device if nationality is salient. We observe that 56.3% of
the subjects (27 out of 48) choose B in treatment S if they know that
their partner has the same nationality. As shown in Figure 1, this
percentage is considerably higher than the equivalent percentage in the
control treatment C (46 out of 108; 42.6%; p = .057, [chi square] =
2.49, one-sided); that is, subjects are approximately 33% more likely to
attempt to coordinate on the Pareto-efficient outcome if they know that
they are matched with a compatriot. This finding is significant at p =
.025 after controlling for the subject's nationality, his/her level
of identity with his/her nationality, GPA, and gender (Table 3, model
1). Identity and GPA predict positively the choice to play B (p <
.062). Model 1 also shows that both Korean and Chinese students tend to
be less likely to choose B in treatment C, as compared to the Japanese
(p <. 144).
Figure 2 illustrates that the strength of the coordination device
(i.e., the information that the partner is a compatriot) is highly
dependent on nationality. We observe the probability that subjects play
B is the highest in the treatment condition S-IN (treatment S, partner
is compatriot) for all three nationalities. However, for Japanese
subjects salient information about the partner being Japanese has very
little impact on the willingness to choose B as compared to having no
information about the partner's nationality in treatment C (50% vs.
47.2%, respectively). By contrast, Korean subjects are more likely to
choose B when paired with another Korean in treatment S than in C (66.7%
vs. 38.9%; p = .047, [chi square] = 2.80, one-sided). Chinese subjects
are more likely to choose B when paired with a compatriot in treatment S
than in C, but this effect is not statistically significant (55.6% vs.
41.7%; p = .167, [chi square] = .93, one-sided).
Figure 3 provides an overview of the outcomes in all treatments and
distinguishes between the PDE outcome, RDE outcome, and MIS. This figure
illustrates that the more pronounced willingness to coordinate on the
PDE leads to more payoff-dominant outcomes (7 out of 24 or 29.2%) when
matched with a compatriot in S than in C (12 out of 54 or 22.2%). It
also leads to less coordination on the RDE than in C (16.7% vs. 37.0%)
and when matched with a non-compatriot in S (31.3%).
How can we explain these coordination game results? Presumably,
greater coordination in S with a compatriot could be explained by either
in-group favoritism or the fact that nationality serves as a
coordination device. We explore subjects' propensity to treat their
compatriots better than non-compatriots by utilizing the results from
ultimatum and dictator games. Interestingly, we do not find that
subjects behave consistently more pro-socially toward their compatriots
in the ultimatum and dictator games in our subject pool. For example, as
reported in Table 2, the mean transfer in the dictator game to a
compatriot in treatment S (3.32) is almost identical to the mean
transfers in treatment C (3.33), and not statistically different from
the mean transfer to a non-compatriot in treatment S (2.83). (6) We find
some differential treatment in the ultimatum game on the side of the
proposer: the mean offer in treatment S to a compatriot (5.25) is higher
than the mean offer in treatment C (4.62, p = .069, Mann-Whitney,
two-sided); however, it is not statistically higher than the mean offer
to a non-compatriot in treatment S (5.06). (7) On the side of the
responder, we find that the minimal acceptable offer is higher with a
compatriot in treatment S than in C, but this difference is not
statistically significant (3.5 vs. 3.13, p = .326, Mann-Whitney test).
One possible explanation for the lack of in-group favoritism in our
subject pool may lie in the dual nature of culture in this experiment.
While nationality certainly comprises one aspect of culture, it does not
represent the full extent of it. Besides being nationals of three
different countries, subjects in this experiment are all students of the
University of Hawaii and are united by a university culture and their
common Asian heritage. As international students, they may perceive each
other as members of the same group, rather than outsiders. This common
culture may explain why we do not find significant in-group favoritism
in the dictator and ultimatum games.
The lack of in-group favoritism suggests that our finding that
compatriots are more willing to coordinate on the PDE is not primarily
driven by a propensity--or a potential experimental demand effect--to
behave more pro-socially toward compatriots.
FINDING 1a: Nationality serves as a coordination device if
nationality is salient: Subjects are more likely to choose a
payoff-dominant equilibrium action if matched with a compatriot. The
strength of this coordination device, however, is highly
nationality-specific.
Next, we assess whether nationality continues to serve as a
coordination device if nationality is non-salient. We observe that 44.1%
(15 out of 34) subjects choose B if they play with someone from the same
nationality in treatment NS. This percentage is not statistically
different from the percentage in the control treatment (42.6%). Note
also that none of the other pieces of information that we gave subjects
about their partner in the NS treatment (age, status, hair color, and
eye color) is significantly related to the choice of B (p > .33,
Spearman). Table 3, Model 2 shows that even after controlling for our
covariates, subjects are not more likely to choose B when matched with a
compatriot in NS (p = .987).
In Figure 3, we observe a low fraction of payoff-dominant outcomes
in treatment NS when paired with a compatriot (12.5%) and a high level
of MIS on the outcomes ((8,1) (1,8)), which is even somewhat higher than
in treatment C (64.7% in NS compared to 40.7% in C; p = .101,
Fisher's exact test, two-sided).
We also do not find that subjects behave more pro-socially toward
their compatriots if nationality is non-salient in the dictator and
ultimatum game. In the dictator game, subjects give 3.35 tokens to
compatriots (treatment C = 3.32). In the ultimatum game, subjects offer
4.28 tokens to compatriots in treatment NS (treatment C = 4.62) and the
minimal acceptable offer is 2.35 when matched with a compatriot in NS
(treatment C = 3.13, p = .185, Mann-Whitney).
FINDING 1b: Nationality does not serve as a coordination device if
nationality is non-salient. Subjects are not more likely to try to
coordinate on the payoff-dominant equilibrium with someone from the same
nationality if they are provided with additional information about their
partner.
The next question we address is whether subjects are less likely to
choose a PDE action if they know that their partner has a different
nationality. We observe that 40.6% (39 out of 96) of the participants
choose B if they play with someone from a different nationality in
treatment S and 42.6% (29 out of 68) of the subjects choose B if they
play with someone from a different nationality in NS. These percentages
are not different from the percentage in treatment C (42.6%, [chi
square], p > .776). The non-significant impact of coming to know that
the partner has a different nationality is also confirmed in Models 3
(for treatment S) and 4 (for treatment NS) of Table 3 (p > .453).
In Figure 3, we can observe that only a low fraction of outcomes
are payoff-dominant (12.5%) when matched with someone from a different
nationality in S. Moreover, 56.3% miscoordinate on the outcomes (8,1) or
(1,8) in S when matched with someone from a different nationality, which
is insignificantly higher than in C (Fisher's exact test, p = .164,
two-sided). Also in treatment NS, few outcomes are payoff-dominant when
matched with someone from a different nationality (14.7%)--but this
percentage is still higher than when matched with someone from the same
nationality in NS (12.5%). More than half of the subjects (55.9%)
miscoordinate on the outcomes (8,1) and (1,8) which is somewhat higher
than in the control treatment (40.7%, p= .192, two-sided, Fisher's
exact test) but less than when matched with someone from the same
nationality in S (64.7%).
FINDING 2: Choice between payoff-dominant and risk-dominant actions
is the same in C and S/NS. That is, knowing that the partner is of
different nationality does not affect the choices in coordination game
as compared to when nationality is unknown.
We conclude this section by investigating the role of the
constellations of nationalities. Figure 4 provides an overview of the
probability of choosing B depending on the treatment and the
constellation of nationalities. The figure shows some interesting,
nationality specific patterns. First, we observe that Japanese subjects,
who seemed, according to Figure 2, not to discriminate between
compatriots and other nationals, discriminate between Koreans (only
27.0% choose B when paired with a Korean in treatment S) and Chinese
(61.0%). Korean subjects tend to be less likely to choose B when the
information about their partner's nationality is non-salient as
compared to when it is salient. The opposite is true for Chinese, who
tend to be less likely to choose B when the information about their
partner's nationality is salient. For example, when a Chinese only
knows about his/her partner that he/she is Japanese (treatment S),
he/she chooses B with a probability of .28 as compared to a probability
of .50 when additional information about the partner is available
besides his/her Japanese nationality (treatment NS).
Interestingly, in S and NS combined subjects are quite unlikely to
choose B if their partner is Japanese and not a compatriot (35.7%). This
percentage is lower than when the partner is Korean (40.4%) or Chinese
(48.2%). While these differences are not uniform and not statistically
significant, the patterns are roughly consistent with a contentious
history of cross-national disagreements in the region. (8) One may
speculate that both the Chinese and Korean subjects are biased against
the Japanese in response to the Japanese imperialist policy and military
occupations of their nations in the past. Recent disputes over the
ownership of the islands in the East China Sea, for example, are
evidence that these regional tensions are not entirely a historical
artifact.
V. CONCLUSION
In this study, we experimentally test the relevance and scope of
focal points when individuals make decisions under strategic
uncertainty. In a period marked by extensive cross-national
interactions, nationality may present an important focal point that
individuals coordinate on. What is the relevance and scope of
nationality as a coordination device? We recruited subjects from three
countries that heavily engage in international trade and let them play
coordination games in three treatments in which we manipulated the
information they receive about their partner. (9)
Our findings suggest that nationality can function as a
coordination device but that the scope of this device is limited.
Subjects attempt to coordinate more on the PDE if their partner has the
same nationality and the information about the partner's
nationality is salient. In addition, our data suggest that the scope of
nationality as a coordination device is highly nationality specific and
provide suggestive evidence showing that this difference is not a result
of more pro-social behavior toward compatriots, that is, in-group
favoritism. However, if the information about the partners'
nationality is not salient, that is, nationality is only one of several
other attributes which subjects know about their partner, nationality
seems to be irrelevant for coordination. Moreover, overall we do not
find in our study that coordination is more difficult between partners
of different nationalities.
One possible explanation of these results may lie in the dual
nature of culture in this experiment. Nationality certainly constitutes
an important part of one's culture and identity, but it does not
represent the full extent of it. While being nationals of three
different countries, the subjects in this experiment are all students of
the University of Hawaii. They are united by a common university culture
and international community. As the focus is diluted away from
nationality in the non-salient treatment, common culture for university
students may dominate and hence the results on nationalities are
weakened. The dual nature of culture in our experiment may also explain
why we do not find significant in-group favoritism in the dictator and
ultimatum games. (10)
More generally, one may speculate that the scope of nationality as
a coordination device is limited because (1) interacting parties often
have access to more information about their partners than only their
nationality and (2) there is a probability that they share different
cultural aspects. On the other hand, our findings also suggest that
having more information about trading partners may not necessarily be
beneficial as it seems to increase the risk of MIS. Moreover, our
experimental results imply that coordination is not more difficult
between parties from different nationalities if subjects have more
information about their partners than only their nationality, as
compared to the benchmark case when nationality is unknown.
These results also contribute to the growing cross-cultural
literature and the economic literature on group identity. Our findings
suggest the necessity of conducting experiments that manipulate the
salience of cultural and group membership for testing the robustness of
the findings in the in- and out-group literature, which use natural
groups. Moreover, our findings point to the need of conducting
additional research to investigate the ultimate mechanism(s) behind the
PDE action. While we have learned through our design that it is not
simply in-group favoritism that drives our nationality coordination
device finding, it would be interesting to better understand the
underlying rationale.
ABBREVIATIONS
GPA: Grade Point Average
MIS: Miscoordination
PDE: Payoff-Dominant Equilibrium
RDE: Risk-Dominant Equilibrium
doi: 10.1111/ecin.12097
Online Early publication May 20, 2014
APPENDIX A. PRE-EXPERIMENTAL QUESTIONNAIRE
1. What is your age? (Mean = 25.01, st. dev = 5.6)
2. Please indicate your gender
--Male (31.19%) --Female (68.81%)
3. What is your major at UH?
4. How long have you lived in the United States? (Mean = 3.71, st.
dev = 4.74; 54.61% of subjects --<2 years)
5. Of which country are you currently a citizen? (Japanese --34.4%,
Korean--30.94%, Chinese--34.62%)
6. How strongly do you identify yourself with this country?
--Not at all (1.7%) --Somewhat (41.51%)
--Very strongly (56.79%)
7. Are you happy to identify yourself with this country?
--No (5.75%) --Somewhat (30.28%)
--Yes (69.72%)
APPENDIX B. EXPERIMENTAL INSTRUCTIONS
TREATMENT C (CONTROL)
Welcome and thank you for your participation in the economics
experiment on decision making!
INTRODUCTION
Please note that you are not allowed to communicate with each other
during the experiment.
During this experiment you will participate in decision tasks that
give you the opportunity to earn money. All the earnings in this
experiment will be in dollars. Immediately upon completion of the
experiment, we will pay you your game earnings in cash. Your earnings
are confidential and you will be paid in private.
This experiment will consist of several parts. In each part, you
will be asked to make three decisions which will involve another
participant with whom you will be randomly matched. For every decision
task, you will be randomly matched with a different participant than in
the previous decision. Your decision may affect the payoffs of others,
just as the decisions of the person you are matched with may affect your
payoffs.
At the end of the experiment, the computer will randomly choose one
task from each part as the paid task. The other tasks will remain
unpaid. You will not be informed of the results of any task until the
end of the experiment.
We will proceed to the decisions once the instructions are clear.
Are there any questions?
PART 1
In this part, you will be randomly matched with one other
participant. Their identity will not be revealed to you and yours will
not be revealed to them.
You will be assigned a role: Proposer or Responder. Your role will
remain the same for all three decisions in this part and in the next.
Instructions to the PROPOSERS: A sum of 10 dollars has been
allocated to both of you. The proposer has to choose how the money
should be divided between you. The proposer's task is to choose an
amount between 0 and 10 dollars to be offered to the responder.
Instructions to the RESPONDERS: The responders will not see the
proposer's offer. The responder's task is to indicate the
smallest amount which they will accept from the proposer. If the
proposer's actual offer to the responder is at least as large as
the smallest offer that responder is willing to accept, then the money
is divided according to the proposer's offer. Otherwise, neither of
you will receive anything.
Your decisions during this experiment will remain anonymous and
private and you will not know the outcome of the decisions until the end
of the experiment. You will be asked to make three decisions in this
part, and each time you will be matched with a different person.
Any questions?
PART 2
In this part, you will be matched with different participants than
before. You have not been matched with these people before. You have
been assigned the same role as in the previous part (proposer or
responder). The proposer has to make a decision, while the responder has
no decision to make in this game. As before, you will not know the
identity of the person you are matched with.
Instructions to the Proposers: A sum of 10 dollars has been
allocated to the both of you. The proposer has to choose how the money
should be divided between you. The proposer's task is to choose an
amount between 0 and 10 dollars to be offered to the responder. The
responder has no decision to make in this game, so the money will be
divided according to the decision of the proposer.
You will be asked to make your decision three times, and each time
you will be matched with a different person.
Please make your decision as prompted on the screen.
PART 3
In this part, you have been randomly matched with another person.
You have not been matched with this person before. Both of you will make
decisions at the same time and your payoff in this part will depend on
your decision as well as the decision of the participant with whom you
are matched. Their identity will not be revealed to you.
Your task in this part is to choose either "A" or
"B."
Depending on your choice and the choice of your match, your
earnings will be the following:
Other participant
A B
A 7,7 8,1
You
B 1,8 9,9
* If you choose "A" and the other participant chooses
"A" then both of you receive $7 each
* If you choose "A" and the other participant chooses
"B" then you receive $8 and the other participant receives $1
* If you choose "B" and the other participant chooses
"B" then both of you receive $9 each
* If you choose "B" and the other participant chooses
"A" then you receive $1 and the other participant receives $8.
Now, please make your decision as prompted on the screen.
APPENDIX C. EXIT QUESTIONNAIRE
1. What is your gender? -- M (31.19%) -- F (68.81%)
2. What is your major at UH? --
3. What is your GPA? -- (Mean = 3.26, st. dev = .69, Max = 4)
4. How long have you been living in the United States? -- (Mean =
3.71, st. dev = 4.74; 54.61% of subjects - less than 2 years)
5. How easy to understand were the instructions? -- -- Difficult
(33.68%) -- Easy (66.32%)
6. [Only applicable to NS treatments] When making decisions in this
experiment, you were matched with another person. Which of their
characteristics were most important in your decision? --
7. Do you have any friends participating in this session at the
same time with you?
--Yes --No
8. If you do have friends in this session with you, did that affect
your decisions in this experiment? -- Yes -- No
9. Did you like the experiment? --Yes --No
10. Please add any additional comments you have about this
experiment:
APPENDIX D
TABLE A1
Published Research with Multi-game Experimental Design
Exact
Implementation
of Multi-game
Design (number
of games,
Place and Year payment
Author(s) of Publication strategy, etc.)
Tomomi, AER, 2010 3 games: trust
Camerer, and game (TG), risk
Nguen and time
discounting. TG
was played
first, but the
outcomes were
not revealed to
the subjects
until the end
of the session.
Subjects were
paid for all 3
games
Chamess, Kami, GEB, 2010 1-4 games:
and Levin transparent
test, public
goods (PG),
winner's curse,
and hidden
information and
communication
experiment.
Some sessions
included
feedback and
others did not
Rode GEB. 2010 6 games/
exercises: 4
math/general
knowledge
contests,
coordination
game or
matching
pennies game,
and
communication
game
Charnes and AER. 2009 3 games: public
Villeval goods,
real-effort
competition
game, and risk
aversion test.
Subjects were
paid for all 3
games. Some
sessions
included
feedback and
others did not
Chen and Li AER, 2009 24 games:
variations of
dictator game
(DG) followed
by 16 response
games (strategy
method). One
random round
was chosen for
payment in each
part. No
outcome was
revealed until
the end of the
experiment
Botelho et al. GEB, 2009 2 games (in
some sessions):
regular PG game
and "sanctions"
PG game (with
the order
reversed in
some sessions)
Corbae and GEB, 2008 5-9 games
Duffy depending on
the session:
variations of
stag-hunt game.
Subjects were
paid for all 3
games
Chamess and QJE,2002 2-8 games
Rabin depending on
the session: 7
variations of
DG and 20
response games.
Subjects who
made 2-8
decisions were
paid for random
1 or 2 choices.
Role reversion
and strategy
method with no
feedback were
used
Henrich et al. AER, 2001 3 games:
ultimatum game
(UG) dictator
game (DG), and
public goods
(PG) game
Anderson, Economic 2 games: UG and
Rodgers, and Letters, 2000 DG (with
Rodriguez reversed order
in some
treatments).
Perfect
stranger
design.
Subjects were
paid for one
randomly chosen
part
Find Order,
Report Results Framing, and
Author(s) for all Games? Other Effects?
Tomomi, No. TG is not No discussion
Camerer, and reported in the of any possible
Nguen study order or
framing effects
Chamess, Kami, No. Only No discussion
and Levin transparent of any possible
test is effects.
discussed
Rode Yes, but the No discussion
analysis of any possible
focuses on the effects
communication
game
Charnes and Yes No discussion
Villeval of any possible
effects, except
the "order"
variable in a
regression is
found to be
insignificant
Chen and Li Yes No discussion
of any possible
effects
Botelho et al. No, the results No discussion
of the of any possible
"sanctions" PG effects
game are not
discussed
Corbae and Yes No discussion
Duffy of any possible
effects
Chamess and Yes No discussion
Rabin of any possible
effects
Henrich et al. Yes No discussion
of any possible
effects
Anderson, Yes Find order
Rodgers, and effects in the
Rodriguez United States,
but not in
Honduras
APPENDIX E. ULTIMATUM AND DICTATOR GAME RESULTS
TABLE A1
Ultimatum and Dictator Game Behavior: Constellation
of Nationalities in NS Treatment
Variable (mean, $) ja-ja ja-ko ja-ch ko-ko ko-ja
Proposer's share 5.4 5.5 5.6 5.6 4.5
In ultimatum game 7 6 5 J 4
Proposer's share 6.0 6.0 5.8 6.6 6.8
In dictator game 6 6 6 5 5
Responder's minimum 1.8 1.8 2.7 2.8 4.2
Acceptable offer 6 4 6 5 6
Variable (mean, $) ko-ch ch-ch ch-ja ch-ko
Proposer's share 6.2 6.2 5.7 6.0
In ultimatum game 5 6 6 6
Proposer's share 5.6 13 6.5 7.3
In dictator game 5 6 6 6
Responder's minimum 3.3 2.5 3.0 2.6
Acceptable offer 6 6 6 5
Notes: Numbers in italics show the respective number of
observations. Ja-ko, for example, shows the mean decision of
the Japanese proposers matched with Korean responders,
whereas ko-ja shows the mean decision of the Korean proposer
matched with a Japanese responder.
ja, Japanese; ch, Chinese; ko, Korean.
TABLE A2
Ultimatum and Dictator Game Behavior: Constellation
of Nationalities in S Treatment
Variable (mean, $) ja-ja ja-ko ja-ch ko-ko ko-ja
Proposer's share 5.0 4.8 4.9 4.7 5.3
In ultimatum game 9# 6# 8# 6# 9#
Proposer's share 6.2 5.8 6.3 6.8 7.9
In dictator game 10# 6# 8# 6# 8#
Responder's minimum 3.3 3.6 3.6 3.2 3.2
Acceptable offer 9# 9# 10# 6# 6#
Variable (mean, $) ko-ch ch-ch ch-ja ch-ko
Proposer's share 5.4 4.6 5.2 3.7
In ultimatum game 8# 9# 10# 7#
Proposer's share 1.4 7.1 7.6 7.6
In dictator game 8# 9# 10# 8#
Responder's minimum 1.9 3.9 3.6 3.5
Acceptable offer 7# 9# 8# 8#
Notes: Numbers in italics show the respective number of
observations. Ja-ko, for example, shows the mean decision of
the Japanese proposers matched with Korean responders,
whereas ko-ja shows the mean decision of the Korean proposer
matched with a Japanese responder.
ja, Japanese; ch. Chinese; ko, Korean.
REFERENCES
Anderson, L., Y. Rodgers, and R. Rodriguez. "Cultural
Differences in Attitudes Toward Bargaining." Economic Letters, 69,
2000, 45-54.
Bornhorst, F., A. Ichino, 0. Kirchkamp, K. Schlag, and E. Winter.
"Trust and Trustworthiness Among Europeans: South-North
Comparison." Mimeo, London, 2008.
Botelho, A., G. Harrison, L. Pinto, and E. Rutstrom. "Testing
Static Game Theory with Dynamic Experiments: A Case Study of Public
Goods." Games and Economic Behavior, 67(3), 2009, 253-65.
Brandts, J., and D. Cooper. "It's What You Say, Not What
You Pay: An Experimental Study of Manager-Employee Relationships in
Overcoming Coordination Failure." Journal of the European Economic
Association, 5(6), 2007, 1223-68.
Brandts, J., S. Tatsuyoshi, and A. Schram. "How Universal Is
Behavior? A Four Country Comparison of Spite and Cooperation in
Voluntary Contribution Mechanisms." Public Choice, 119, 2004,
381-424.
Buchan, N., E. Johnson, and R. Croson. "Let's Get
Personal: An International Examination of the Influence of
Communication, Culture and Social Distance on Other Regarding
Preferences." Journal of Economic Behavior and Organization, 60(3),
2006, 373-98.
Buchan, N., G. Grimalda, R. Wilson, M. Brewer, E. Fatas, and M.
Foddy. "Globalization and Human Cooperation." Proceedings of
the National Academy of Science, 106(11), 2009, 4138-42.
Camerer, C. Behavioral Game Theory: Experiments in Strategic
Interaction. Princeton, NJ: Princeton University Press, 2003.
Charness, G. "Self-serving Cheap Talk and Credibility: A Test
of Aumann's Conjecture." Games and Economic Behavior, 33,
2000, 177-94.
Charness, G., and M. Rabin. "Understanding Social Preferences
with Simple Tests." Quarterly Journal of Economics, 117(3), 2002,
817-69.
Charness, G. R., and A. Rustichini. "Individual Behavior and
Group Membership." American Economic Review, 97, 2007, 1340-52.
Charness, G., and M. Villeval. "Cooperation and Competition in
Intergenerational Experiments in the Field and the Laboratory."
American Economic Review, 99(3), 2009. 956-71.
Charness, G., E. Kami, and D. Levin. "On the Conjunction
Fallacy in Probability Judgment: New Experimental Evidence Regarding
Linda." Games and Economic Behavior, 68, 2010, 551-56.
Chen, R., and Y. Chen. "The Potential of Social Identity for
Equilibrium Selection." American Economic Review, 101(6), 2011,
2562-89.
Chen, Y., and X. Li. "Group Identity and Social
Preferences." American Economic Review, 99(1), 2009, 431-57.
Chuah, S., R. Hoffman, M. Jones, and G. Williams. "Do Cultures
Clash? Evidence from Cross-National Ultimatum Game Experiments."
Journal of Economic Behavior and Organization, 64, 2007, 35-48.
Cooper, R., D. Dejong, R. Forsythe, and T. Ross. "Selection
Criteria in Coordination Games: Some Experimental Results."
American Economic Review, 80(1), 1990, 218-33.
Corbae, D., and J. Duffy. "Experiments with Network
Formation." Games and Economic Behavior, 64, 2008, 81-120.
Crawford, V. "Adaptive Dynamics in Coordination Games."
Econometrica, 63(1), 1995, 103-43.
Crawford, V., and H. Haller. "Learning How to Cooperate:
Optimal Play in Repeated Coordination Games." Econometrica, 58(3),
1990, 571-95.
Crawford, V., U. Gneezy, and Y. Rottenstreich. "The Power of
Focal Points Is Limited: Even Minute Payoff Asymmetry May Yield Large
Coordination Failures." American Economic Review, 98(4), 2008,
1443-58.
Eckel, C.. and P. J. Grossman. "Managing Diversity by Creating
Team Identity." Journal of Economic Behavior and Organization,
58(3), 2005, 371-92.
Eckel, C., and P. J. Grossman. "Subsidizing Charitable
Contributions in the Field: Evidence from a Non-Secular Charity."
Mimeo, Dallas, TX, 2006.
Efferson, C., R. Lalive, and E. Fehr. "The Evolution of
Cultural Groups and Ingroup Favoritism." Science, 321, 2008,
1844-49.
Fershtman, C., and U. Gneezy. "Discrimination in a Segmented
Society: An Experimental Approach." Quarterly Journal of Economics,
116(1), 2001, 351-77.
Fischbacher, U. "z-Tree: Zurich Toolbox for Ready-Made
Economic Experiments." Experimental Economics, 10(1), 2007, 171-78.
Goette, L., D. Huffman, and S. Meier. "The Impact of Group
Membership on Cooperation and Norm Enforcement: Evidence Using Random
Assignment to Real Social Groups." American Economic Review, 96(2),
2006, 212-16.
der Heijden, V., J. Nelissen, J. Potters, and H. Verbon.
"Simple and Complex Gift Exchange in the Laboratory." Economic
Inquiry, 39, 2001, 280-97.
Henrich, J. "Does Culture Matter in Economic Behavior?
Ultimatum Game Bargaining Among the Machiguenga." American Economic
Review, 90, 2000, 973-79.
Henrich, J., R. Boyd, S. Bowles, C. Camerer, H. Gintis, R.
McElreath, and E. Fehr, "In Search of Homo Economicus: Experiments
in 15 Small-Scale Societies." American Economic Review, 91, 2001,
73-79.
Holm, H. "Gender-based Focal Points." Games and Economic
Behavior, 32(2), 2000, 292-314.
Kollock, P. "Transforming Social Dilemmas: Group Identity and
Co-operation, " in Modeling Rationality, Morality, and Evolution,
edited by P. A. Danielson. Oxford: Oxford University Press, 1998,
186-210.
Kroeber, A. L., and C. Kluckhorn. "Culture: A Critical Review
of Concepts and Definitions." Papers of the Peabody Museum 47,
1952.
Kuwabara, K., R. Wilier, M. Macy, R. Mashima, S. Terai, and T.
Yamagishi. "Culture. Identity, and Structure in Social Exchange: A
Web-Based Trust Experiment in the United States and Japan." Social
Psychology Quarterly, 70(4), 2007, 461-479.
McLeish. K., and R. Oxoby. "Identity, Cooperation and
Punishment." Mimeo, Germany, 2007.
Mehta, J., C. Starmer, and R. Sugden. "The Nature of Salience:
An Experimental Investigation of Pure Coordination Games." American
Economic Review, 84(3), 1994, 658-73.
Okada, A., and A. Riedl. "When Culture Does Not Matter:
Experimental Evidence from Coalition Formation in Ultimatum Games in
Austria and Japan." Working Paper, University of Amsterdam (CREED),
1999.
Rode. J. "Truth and Trust in Communication: Experiments on the
Effect of a Competitive Context." Games and Economic Behavior,
68(1), 2010, 325-38.
Roth. A. E., M. Prasnikar, V. Okuno-Fujiwara, and S. Zamir.
"Bargaining and Market Behavior in Jerusalem, Ljubljana, Pittsburgh
and Tokyo: An Experimental Study." American Economic Review, 81,
1991, 1068-109.
Schelling, T. The Strategy of Conflict. Cambridge, M A: Harvard
University Press, 1960.
Sherif, M., O. Harvey, B. J. White, W. R. Hood, and C. Sherif.
Intergroup Conflict and Co-operation: The Robbers Cave Experiment.
Norman: University of Oklahoma, 1961.
Sugden, R. The Economics of Cooperation, Rights and Welfare.
Oxford: Basil Blackwell, 1986.
--. "A Theory of Focal Points." The Economic Journal,
105, 1995, 533-50.
Tajfel, H., and J. Turner "An Integrative Theory of Intergroup
Conflict," in The Psychology of Intergroup Relations, edited by W.
G. Austin, and S. Worchel. Monterey, CA: Nelson-Hall, 1979.
Takshashi, C., T. Yamagishi, J. Liu, F. Wang, Y. Liin, and S. Yu.
"The Intercultural Trust Paradigm: Studying Joint Cultural
Interaction and Social Exchange in Real Time over the Internet."
International Journal of Intercultural Relations, 32(3), 2008, 215-28.
Tomomi, T., C. Camerer, and T. Nguyen. "Risk and Time
Preferences: Linking Experimental and Household Survey Data from
Vietnam." American Economic Review, 100, 2010, 557-71.
Van Huyck, J., R. Battalio, and R. Beil. "Tacit Coordination
Games, Strategic Uncertainty, and Coordination Failure." American
Economic Review, 80, 1990, 234-48.
Yamagishi, T., Y. Makimura, M. Foddy, M. Matsuda, T. Kiyonari, and
M. Platow. "Comparisons of Australians and Japanese on Group-Based
Cooperation." Asian Journal of Social Psychology, 8(2), 2005,
173-90.
Yamagishi, T., H. Hashimoto, and J. Schug. "Preferences versus
Strategies as Explanations for Culture-Specific Behavior."
Psychological Science, 19(6), 2008, 579-84.
(1.) We use the term culture, cultural background, and nationality
interchangeably in this study. For a survey on the different definitions
for culture, consult, for example, Kroeber and Kluckhorn (1952). Culture
certainly has many meanings and nationality is just one aspect of it,
albeit an important one. We use nationality as a proxy for culture and
discuss the strength of this proxy and other aspects of culture in the
conclusion.
(2.) In principle, there could be order effects, that is, subjects
may in general play a coordination game differently after playing, for
example, a dictator game. However, because our main analysis compares
treatments where the order was identical, our treatment differences
cannot be attributed to order effects. We briefly examine relationships
between ultimatum game, dictator game, and coordination game behavior in
Section III.
(3.) The necessity to conduct multiple games in one session was
motivated mainly by the unique subject pool in this experiment. The
nature of the experiment required an equal number of subjects of three
different nationalities to be present in one session at a time. It would
be extremely difficult to find a sufficient number of participants from
different nationalities if we had to avoid a multi-game design. Our
multi-game design was also motivated by the many recently published
experimental studies using multi-game designs (see Appendix D).
(4.) An extra session for the S treatment had to be conducted as a
result of an insufficient number of Korean participants present at one
of the previous sessions.
(5.) Subjects were recruited with campus flyers and email
announcements. Interested individuals were asked to fill out
sociodemographic information as well as to report their nationality. We
invited only individuals who reported to be Japanese, Korean, or
Chinese. However, during the experimental sessions four participants
answered in the pre-experimental questionnaire to be of another
nationality (one participant in NS and three participants in S).
Accordingly, in our analysis we exclude these participants and the
participants that were matched with these in the S and NS treatments.
More precisely, we excluded from our analysis three observations in NS
and nine observations in S. None of the subjects analyzed in our results
had a dual nationality. Upon entering the laboratory, subjects were not
allowed to talk to each other.
(6.) Note also that if we only look at Koreans--in whom the
propensity to coordinate with a compatriot is most pronounced--we also
find no such statistical differences. In the dictator game, Koreans even
give more in treatment C (3.73) than to compatriots in treatment S
(3.17).
(7.) This finding is in line with the finding of Chuah et al.
(2007) who find that Malaysian Chinese subjects gave on average higher
offers in the ultimatum game to compatriots as compared to subjects from
the United Kingdom. Note that
there are procedural differences between our and their study. In
particular, Chuah et al. conducted their experiments in their
subjects' respective home countries (Malaysia or United Kingdom)
whereas we conducted our experiments outside the subjects' home
countries.
(8.) Kuwabara et al. (2007) and Takshashi et al. (2008) find
similar nationality specific effects in their variant of trust game
between participants from Japan, China, and Taiwan and suggest that
culture-specific content (e.g., collective guilt for WWII) may influence
these national-level patterns.
(9.) According to the size of their exports China ranks 2nd, Japan
4th, and Korea 12th. From 2000 to 2008, China has almost sixfolded,
Japan and Korea have more than doubled, their exports. Data from United
Nations Statistics Department available at http://unstats.
un.org/unsd/trade/imts/annual%20totals.htm.
(10.) Comparably, Buchan et al. (2009) find that globalization
promotes cooperation.
OLGA STODDARD, and ANDREAS LEIBBRANDT *
* We would like to thank Katerina Sherstyuk for her helpful
feedback on the paper, as well as Heber Moulton for his excellent
research assistance, and participants of the 2012 ESA conference in
Tucson, AZ and the research seminars at University of Hawaii and Brigham
Young University for their helpful comments.
Stoddard: Visiting Assistant Professor, Department of Economics,
Brigham Young University, 149 FOB Provo, UT 84606. Phone 1-801-422-3580,
Fax 1-801-422-0194, Email olga.bogach@byu.edu
Leibbrandt: Assistant Professor, Department of Economics, Monash
University, Clayton, Victoria 3800. Phone 61-3-9905-2478, Fax
61-3-9905-5476, E-mail andreas. leibbrandt@monash.edu
TABLE 1
Experimental Design (number of observations
by treatment and matching)
Salient Non-salient
Control (information (information about
(no information about nationality
about nationality and other
the match) only) characteristics)
J-J 11 18 12
K-K 11 12 10
C-C 11 18 12
J-K 21 30 22
J-C 22 36 24
K-C 22 30 22
Total 108 144 102
Notes: J-K represents a matching in which a Japanese subject
is paired with a Korean subject. J, Japanese; K, Korean;
C, Chinese.
TABLE 2
Behavior in Coordination, Ultimatum and Dictator Games
(means, number of observations in parentheses)
S-Treatment
C- Non-
Treatment compatriot Compatriot
Coordination 0.43 0.41 0.56
(probability B) (108) (96) (48)
Ultimatum 4.62 5.06 5.25
(proposer offer) (53) (48) (24)
Ultimatum 3.13 3.27 3.50
(responder MAO) (53) (48) (24)
Dictator 3.33 2.83 3.32
(proposer offer) (54) (48) (25)
NS-Treatment
Non-
compatriot Compatriot
Coordination 0.43 0.44
(probability B) (68) (34)
Ultimatum 4.38 4.28
(proposer offer) (32) (18)
Ultimatum 3.00 2.35
(responder MAO) (33) (17)
Dictator 3.65 3.35
(proposer offer) (34) (17)
Notes: Bold numbers indicate statistically significant
differences. MAO, minimum acceptable offer as stated by a
responder in ultimatum game.
TABLE 3
Choice of Pareto-dominant Equilibrium Action in Coordination
Game Depending on Matching and Covariates (Probit regression)
(1) (2)
Model C & S C & NS if
Observations in if paired paired w/
treatments w/compatriot compatriot
Compatriot? 0.055 ** 0.001
(0.025) (0.987)
Non-compatriot?
Identity 0.207 * 0.074
(0.062) (0.556)
Female? 0.090 0.097
(0.434) (0.459)
GPA 0.276 *** 0.121
(0.000) (0.216)
Subject is Korean? -0.227 -0.130
(0.126) (0.394)
Subject is Chinese? -0.233 -0.085
(0.102) (0.563)
Number of 154 142
observations
(3) (4)
Model C & S if C & NS if
Observations in paired w/ paired w/
treatments non-compatriot non-compatriot
Compatriot?
Non-compatriot? 0.025 -0.012
(0.448) (0.912)
Identity 0.231 ** 0.061
(0.012) (0.588)
Female? 0.029 0.149
(0.770) (0.213)
GPA 0.091 0.088
(0.199) (0.345)
Subject is Korean? -0.153 -0.190
(0.224) (0.157)
Subject is Chinese? -0.171 -0.060
(0.163) (0.655)
Number of 200 176
observations
Notes: Coefficients are average marginal effects, p values
in parentheses. Robust standard errors clustered on subject
level. Compatriot = 1 if subject is paired with compatriot,
0 otherwise. Non/compatriot = 1 if subject is paired with
subject having a different nationality, 0 otherwise. Subject
is Korean/Chinese? 1 if yes, 0 otherwise. Female = 2 if
subject is a female, 1 otherwise.
* p < .1, ** p < .05, *** p < .01.
FIGURE 1
Choice in Coordination Game Depending on Treatment and Match
A B
Control 54.4 42.6
S: same nationality 43.8 56.2
S: different nationality 59.4 40.6
NS: same nationality 55.9 44.1
NS: different nationality 57.4 42.6
Notes: S, salient treatment; NS, non-salient treatment.
Notes: Table made from bar graph.
FIGURE 2
Choice of Pareto-Dominant Equilibrium Action Depending on Nationality
and Treatment
Japanese Korean Chinese
C 47.2 38.9 41.7
S-IN 50.0 66.7 55.6
S-OUT 45.5 43.3 33.3
NS-IN 41.7 40.0 50.0
NS-OUT 47.8 27.3 52.2
Notes: C: control treatment, subject does not know nationality of
partner; S-IN: Treatment S, subject is paired with compatriot;
S-OUT: Treatment S, subject is paired with partner from a different
nationality; NS-IN: Treatment NS, subject is paired with
compatriot, NS-OUT: Treatment NS, subject is paired with partner
from a different nationality.
Notes: Table made from bar graph.
FIGURE 3
Outcomes in Coordination Game Depending on Treatment and Match
Outcome in coordination game
PDE RDE MIS
Control 22.2 37.0 40.7
S: same nationality 29.2 16.7 54.2
S: different nationality 12.5 31.2 56.2
NS: same nationality 11.8 23.5 64.7
NS: different nationality 14.7 29.4 55.9
Notes: PDE, payoff-dominant equilibrium outcome with payoffs (9,9);
RDE, risk-dominant equilibrium outcome with payoffs (7,7); MIS,
miscoordination outcome with payoffs (1,8) or (8,1). S, salient
treatment; NS, non-salient treatment.
Notes: Table made from bar graph.
FIGURE 4
Choice of Pareto-dominant Equilibrium Action Depending on Treatment
and Constellations of Nationalities
Japanese Korean Chinese
S-JK 26.7 40.0 27.8
S-JC 61.1 46.7 40.0
NS-JK 45.5 27.3 50.0
NS-JC 50.0 27.3 54.6
Notes: S-JK: Treatment S, Japanese subject is paired with Korean;
S-JC: Treatment S, Japanese subject is paired with Chinese; NS-JK:
Treatment NS, Japanese subject is paired with Korean; NS-JC:
Treatment S, Japanese subject is paired with Chinese; S-KJ:
Treatment S, Korean subject is paired with Japanese; S-KC:
Treatment S, Korean subject is paired with Chinese; NSKJ: Treatment
NS, Korean subject is paired with Japanese; NS-KC: Treatment NS,
Korean subject is paired with Chinese; S-CJ: Treatment S, Chinese
subject is paired with Japanese; S-CK: Treatment S, Chinese subject
is paired with Korean; NS-CJ: Treatment NS, Chinese subject is
paired with Japanese; NS-CK: Treatment NS, Chinese subject is
paired with Korean.
Notes: Table made from bar graph.
