DISCRIMINATIONINASEGMENTEDSOCIETY: …rady.ucsd.edu/faculty/directory/gneezy/pub/docs/... · 2012. 7. 5. · nationidenti”edinourtrustgameexperimentwasonlyapparent withrespecttoandbymaleplayers

DISCRIMINATION IN A SEGMENTED SOCIETYAN EXPERIMENTAL APPROACH

CHAIM FERSHTMAN AND URI GNEEZY

This paper proposes an experimental approach to studying different aspectsof discrimination We let participants play various games with opponents ofdistinct ethnic afliation Strategies based upon such ethnic afliation providedirect evidence of ethnic discrimination This approach was utilized to studyethnic discrimination in Israeli Jewish society Using the ldquotrust gamerdquo we de-tected a systematic mistrust toward men of Eastern origin A ldquodictator gamerdquoexperiment indicated that this discrimination was due to (mistaken) ethnic ste-reotypes and not to a ldquotaste for discriminationrdquo The ldquoultimatum gamerdquo enabled usto trace another ethnic stereotype that reversed the discriminationrsquos directionOne of the surprising results is that this ethnic discrimination is an entirely malephenomenon

I INTRODUCTION

Ethnic discrimination like any other type of discriminationcan be the outcome of a ldquotaste for discriminationrdquo or the result ofethnic stereotypes A taste for discrimination describes a situa-tion wherein individuals are willing to sacrice money wages orprots in order to cater to their prejudice (see Becker [19571993]) In this case the prejudice is already part of the utilityfunction and may reect some dislike anger or similar emotionstoward a certain group of people Ethnic stereotypes on the otherhand are a simplied and standardized perception of a person ora group commonly held by people1 This perception may affect theinteraction between individuals as it may affect the beliefs re-garding the plausible actions or abilities of members of certaingroups Stereotypes are not one-dimensional as they may involvebeliefs pertaining to various types of characteristics and abilitiesSome stereotypes may be statistically correct while others may betotally mistaken

Empirical evidence of discrimination in different societies

We thank Gary Bornstein Yinon Cohen Werner Guth Itzhak HaberfeldErnan Harovy Rachel Croson Moshe Semyonov seminar participants at theUniversity of Haifa Tel Aviv University Technion Humboldt University of Ber-lin Bonn University and Stockholm University as well as the Editor and tworeferees for their helpful comments and suggestions We also wish to thank TheYitzhak Rabin Center for Israel Studies the Sapir Center for Development andthe Israeli Ministry of Science for their nancial support

1 Clearly there are situations in which the two types of prejudices areclosely related and the ethnic stereotype leads to preferences with a taste fordiscrimination

copy 2001 by the President and Fellows of Harvard College and the Massachusetts Institute ofTechnologyThe Quarterly Journal of Economics February 2001

351

has been extensively discussed in the literature2 However whilethese studies are useful in providing evidence for discriminationin different markets they are less so when it comes to explainingthe type of discrimination identied In this paper we propose anexperimental approach to studying the different aspects of dis-crimination In these experiments individuals played a variety ofgames against players of different ethnic groups We subse-quently examined how the ethnic afliation of game partnersaffected the strategy chosen by the players The outcome of theseexperiments may clarify the following questions

(i) Is there discrimination (differential treatment) based onethnic afliation

(ii) Does the discrimination reect a group bias in that eachplayer favors players of his own group or is there asystematic discrimination against one or several ethnicgroups

(iii) Is this discrimination based on a taste for discrimina-tion or is it the outcome of ethnic stereotyping thataffects the playersrsquo assessment regarding their gamepartnersrsquo strategic responses or relevant characteristics

(iv) Are the ethnic stereotypes accurateThis experimental approach was used to examine ethnic

discrimination within Israeli Jewish society3 The latter is char-acterized by an ethnic structure based primarily on country oforigin The two major ethnic groups are Ashkenazic Jews (Euro-pean and American immigrants and their Israeli-born offspring)and Eastern Jews (Asian and African immigrants and their Is-raeli-born offspring) There are persistent economic gaps betweenthe two ethnic segments Ashkenazic immigrants achieve higherlevels of education and earnings than do Eastern immigrantsand these gaps continue to prevail among second-generation im-migrants (see Cohen and Haberfeld [1998] Eisenstadt [1985]

2 There is ample literature on audit studies showing direct evidence ofdiscrimination in employment housing and credit (see for example Fix andStruyk [1993] as well as discussion and references on the subject in Altonji andBlank [1999] and Neumark [1996]) See also the recent special Symposium issueon Discrimination in Product Credit and Labor Markets that appeared in theJournal of Economic Perspectives [1998] and the following papers in this sympo-sium Yinger [1998] Ladd [1998] Darity and Mason [1998] Arrow [1998] Heck-man [1998] and Loury [1998]

3 A clearly signicant and heated segmentation in Israeli society is thatbetween Arabs and Jews but we reserve this issue for future research Anothermajor segmentation in Israeli Jewish society is the one between secular andOrthodox Jews

352 QUARTERLY JOURNAL OF ECONOMICS

Mark [1994] Semyonov and Kraus [1983] Semyonov and Le-renthal [1991] and Shavit [1984])

In order to address question (i) we rst conducted an exper-iment with the ldquotrust gamerdquo (see Berg Dickhaut and McCabe[1995]) We started by examining trust since we believe thateconomic interactions are not governed solely by contractualagreements and that trust between individuals plays an impor-tant role in facilitating efcient activities As Arrow [1972]pointed out ldquovirtually every commercial transaction has withinitself an element of trustrdquo The trust game is a two-player game inwhich Player A is given a xed amount of money and asked todecide whether to transfer any of it to Player B and if so howmuch The experimenter then triples the amount and gives it toPlayer B who is asked to choose whether to transfer any moneyback to Player A In such a game gains are obtainable throughcooperation The game is labeled a trust game as the amount thatPlayer A transfers to Player B serves as an indication of his trustin his game partner or of the two playersrsquo ability to cooperate Theefcient outcome which maximizes the total pie would requirePlayer A to transfer all of his resources to Player B (as theseresources would then be tripled) The subgame perfect equilib-rium on the other hand implies no transfers and thus does notexploit the potential gains deriving from transfer Berg Dick-haut and McCabe [1995] found that typically Player A trans-ferred a positive amount of money to Player B who often returnedan even larger amount4

Our trust game experiment indicates that the segmentedstructure of Israeli society indeed manifests itself in a consistentpattern of discrimination and mistrust The amount of moneytransferred to players of Eastern origin was signicantly lowerthan that transferred to players of Ashkenazic origin

With regard to question (ii) we found no evidence for a groupbias The systematic mistrust of players of Eastern origin wascommon not only among Ashkenazic players but also amongEastern players who themselves discriminate against playersfrom their own group

Discrimination in the trust game however does not neces-sarily indicate mistrust as it could be the outcome of a taste for

4 A similar procedure was used by Camerer and Weigelt [1988] FehrKirchsteiger and Riedl [1993] Fehr Gachter and Kirchsteiger [1997] GuthOckenfels and Wendel [1994] and McKelvey and Palfrey [1992]

353DISCRIMINATION IN A SEGMENTED SOCIETY

discrimination (see question (iii)) The lower amounts transferredto players of Eastern origin could be a reection of prejudicestemming from the playersrsquo preferences rather than from ethnicstereotypes with respect to trust5 We therefore conducted a sec-ond experiment in which the trust game was played while omit-ting the gamersquos second stage This is a one-stage game in whichPlayer A decides on a division of the amount of money given tohim between himself and Player B (the experimenter automati-cally tripled any amount transferred to Player B) This game isknown as the ldquodictator gamerdquo Player B in this game does nothave any strategic role and thus ethnic stereotypes which mayprovide signals regarding his strategic behavior during the sec-ond stage of the game have no bearing on this game Thereforeany transfer distribution differences in the dictator game must bedue to a taste for discrimination We conducted the dictator gameexperiment in the same fashion and with the same population asthat used for the trust game Although the transfer distributionsto Eastern and Ashkenazic players were somewhat differentin this experiment on average the two groups received simi-lar transfers Thus by comparing the transfers in the trustgame with those of the dictator game we can conclude thatethnic discrimination in the trust game is indeed the outcomeof ethnic stereotypes rather than a reection of a ldquotaste fordiscriminationrdquo6

The discrimination that we identied in the trust and ulti-matum games may have a market-based explanation if indeedplayers of different ethnic background play these games differ-ently In this case the playersrsquo ethnic afliation provides a valu-able signal regarding their future choice of action This type ofdiscrimination is denoted in the literature as ldquostatistical discrim-inationrdquo (see Arrow [1973 1998]) In order to check for statisticaldiscrimination in our experiments we examined the responses ofstudents who played the role of Player B in both the trust andultimatum games We found no evidence for such statistical dis-crimination The strategic choice of players of different ethnic

5 Note however that this explanation is less likely in this case as we foundno difference in the behavior of Ashkenazic and Eastern players

6 It is interesting to contrast the results reported in this paper with ex-periments that we conducted concerning the relationship between secular andreligious Jews in Israel [Fershtman and Gneezy 2000] In this experiment thetransfers of secular players to religious partners were lower than the transfers tosecular players both in the trust and dictator games Thus in this case there isa taste for discrimination


backgrounds was not statistically different This observationhowever does not rule out rational behavior Players may havemisperceptions regarding the behavior of players from differentethnic groups and this misperception may rationally lead to adifferent strategy choice We may thus conclude that while ethnicdiscrimination in the trust game in Israeli society is based onethnic stereotypes these are largely mistaken stereotypes

Trust is not the only ethnic stereotype that can be studied bymeans of experiments By changing the game again we were ableto trace another type of ethnic stereotype In the third experimentreported in this paper once again we changed the last stage ofthe game such that Player Brsquos role consisted of just saying ldquoyesrdquoor ldquonordquo to Player Arsquos proposal If Player B chose ldquoyesrdquo the proposalwas implemented otherwise both got zero payoffs This game isknown as the ldquoultimatum gamerdquo and has been extensively dis-cussed in the experimental economics literature7 The commonresult in the ultimatum game experiment is that the proposer(Player A) typically transfers a nonnegligible amount to the re-sponder (Player B) who occasionally rejects low offers The non-negligible transfer by the proposers can be partially explained bytheir wish to achieve a more equitable division of funds but alsoby their belief that the responder may reject ldquounfairrdquo divisionsOur ndings showed that the direction of the discriminationidentied in the trust game was reversed in the ultimatum gamePlayers of Eastern origin received larger transfers than players ofAshkenazic origin This result may reect a familiar ethnic ste-reotype in Israel whereby people of Eastern ethnic backgroundtend to be more driven by a sense of ldquohonorrdquo or to react harshlywhen treated unfairly Given such ethnic stereotypes playerstend to be more careful with proposing low offers to players ofEastern origin for fear that such offers may be rejected

The discrimination pattern may depend not only on ethnicafliation but also on gender We were surprised to nd that theethnic discrimination pattern identied in the trust game was aprimarily male phenomenon Women in the trust game experi-ment transferred similar amounts of money to partners of East-ern or Ashkenazic background and there was no evidence ofethnic discrimination between women8 Thus the ethnic discrimi-

7 See the original experiment by Guth Schmittberger and Schwarze [1982]and the surveys in Camerer and Thaler [1995] Guth [1995] and Roth [1995]

8 Gender-dependent behavior comparisons have a long tradition in the social


nation identied in our trust game experiment was only apparentwith respect to and by male players

II A SHORT BACKGROUND OF EASTERN AND ASHKENAZIC ETHNIC

GROUPS IN ISRAEL

When the State of Israel was established in 1948 there were600000 Jews in Israel primarily of Ashkenazic origin From1948 to 1952 the countryrsquos Jewish population more than doubledas a result of a massive immigration of 700000 Jews About halfof these immigrants were Ashkenazic (Holocaust survivors) theother half consisted of Eastern Jews Immigration continued after1952 but at a slower pace However 55 percent of the immi-grants during the slow period were Eastern Jews The ethnicdivision between the immigrants changed dramatically in the1990s with the massive immigration of Jews from the formerSoviet Union

The persistent social and economic gaps between the twoethnic groups have been the subject of extensive research (egAmir [1987] Cohen and Haberfeld [1998] Haberfeld [1992]Mark [1994] Semyonov and Kraus [1983] and Semyonov andLerenthal [1991]) These studiesrsquo main nding is that the gapshave not been bridged over time Moreover the (education andearning) gaps between second-generation immigrants are nosmaller than those identied among rst-generation immigrantsIn these studies the earning gaps are mainly attributed to school-ing gaps and not to discrimination in the labor market Anexamination of several descriptive statistics of second-generationJewish immigrants in Israel (see Cohen and Haberfeld [1998])reveals that in 1992 for instance the percentage of men ofAshkenazic origin with a college degree (or higher) was 414percent while for men of Eastern origin the gure was 111percent (the corresponding percentages for women were verysimilar) The mean annual earning of men of Eastern origin thatyear was only 677 percent of that of Ashkenazic men (amongwomen the gap was about 20 percent)

sciences See for example Bolton and Katok [1995] Andreoni and Vesterlund[2001] Croson and Buchan [1999] and Eckel and Grossman [1997] and for arecent survey of this literature Eckel and Grossman [1998] This literaturersquos mainnding is that men are more selsh than women These ndings however are farless conclusive and appear to depend heavily on the context of the study


III TRUST AND COOPERATION

Intuition suggests that the degree of trust between individ-uals may play an important role in societyrsquos development andeconomic success Indeed in a multicountry comparison Knackand Keefer [1997] showed that trust was associated with strongereconomic performance These ndings are not surprising if onethinks of the role of trust in promoting business partnershipsbetween rms joint research ventures etc

We have adopted Berg Dickhaut and McCabersquos [1995] trustgame experiment in order to examine trust between differentethnic groups in Israel At the rst stage of this two-player gamePlayer A receives a xed amount of money that he needs toallocate between himself and Player B The experimenter triplesthe amount transferred to Player B and at the second stagePlayer B decides how much of this amount to transfer back toPlayer A

Experimental Procedure

The participants in this experiment consisted of 966 Israeliundergraduates All participants were selected from large (atleast 70 students per class) mandatory courses The rst group ofplayers consisted of students from the University of Haifa and theAcademic College of Tel Aviv Using name lists we chose stu-dents from this group with typical ethnic names9 122 Ashkenazicmale names 135 Eastern male names 118 Ashkenazic femalenames and 108 Eastern female names (483 in total)10 Thesestudents played the role of Player B The second group of playersconsisted of undergraduates from Tel Aviv University Thesestudents recruited in their classes played the role of Player A Inorder to ensure against bias students were randomly matchedand the same experimenter conducted all sessions

The instructions are presented in Appendix 1 The studentswho assumed the role of Player A were told that the experimentwas being conducted in pairs and that they had already beenmatched with another student from another university Thename of the person with whom they were matched was written (inink) at the bottom of the sheet of paper they received

9 Many of the family names in Israel provide a good indication of ethnicafliation a fact well recognized by most of the population

10 See Holm [2000] for an experimental procedure using names to signalgender


Player A was then told that (s)he would receive NIS 2011 andthat hisher partner (Player B) would not receive any moneyPlayer A was asked to decide whether (s)he wanted to transferany portion of the NIS 20 to Player B and if so how much Theplayers were told that the amount transferred would automati-cally be tripled and that Player B with whom they were matchedwould be informed within a few days about all details pertainingto the game including the amount that Player A transferred tohimher Player B would then be asked to decide whether (s)hewanted to send any portion of the money (s)he received back toPlayer A The students were told that this last transfer wouldconclude the experiment and that we would come to their classesone week later to pay them Player A students were then asked towrite down their names and the amount they wanted to transferto Player B

In the second stage of the experiment we approached thestudents from the University of Haifa and the Academic Collegeof Tel Aviv in their classes We told them that we had used the listof names of course participants and that only some of the stu-dents were needed for the experiment (although they were notinformed of the rule used to select the names) The names werethen called out loud and the forms handed to the respectivestudents The students assuming the role of Player B were pro-vided with exactly the same description of the experiment Theywere also informed of the names of Player A with whom they werematched as well as of the sum (s)he had decided to forward themOnce they decided upon the amount they wished to return theywere paid condentially and in cash

After making hisher choice each participant was asked toll out a questionnaire The questions we were most interested inwere the participantrsquos gender and the birthplace of hisher par-ents These types of questions are not uncommon in ofcial formsin Israel Table I presents the number of pairs that were matchedaccording to gender and ethnicity12

It should be emphasized that in our procedure each participantwas aware of the name of the person with whom (s)he was matchedand that the participantsrsquo decisions were revealed to the experi-menter at the end of the experiment This procedure is contrary to

11 At the time of the experiment $1 = NIS 3612 We placed all the students we were unable to classify as Ashkenazic or

Eastern in the group labeled ldquoIsraelrdquo (for example students with parents fromboth groups)


that of Berg Dickhaut and McCabe [1995] who used a doubleanonymity procedure to ensure that no one aside from the partici-pant himselfherself would be aware of hisher decision

Segmented Society

Our rst task was to verify the casual observation wherebyIsraeli society is indeed segmented along ethnic lines not onlywith respect to customs and culture but more importantly alsowith respect to trust Figure I describes the distribution of moneytransfers by Player A to males playing the role of Player Baccording to the ethnic identity of Player B (As evidenced lateron trust is dependent upon gender as well as ethnic afliation)The dark and light columns respectively illustrate the distribu-tion of transfers when Player A was matched with a game partnerwith a typically Eastern (or respectively Ashkenazic) name Thedifference between the two distributions was striking In par-ticular almost 60 percent of the students chose the efcienttransfer (transferring the whole pie) when their opponent was ofAshkenazic origin but only 20 percent did so when they faced anopponent of Eastern origin Figure I tells us the whole (sad) storyof social segmentation in Israeli societyObservation 1 Israeli society is (strongly) segmented In thetrust game the average transfer to an Ashkenazic male partnerwas 1515 whereas the average amount transferred to an Easternmale partner was 806 (or about 53 percent of the average trans-

TABLE IPARTICIPATING PAIRS IN THE TRUST GAME ACCORDING TO GENDER AND ETHNICITY

Player A

Israelimale

Ashkenazicmale

Easternmale

Israelifemale

Ashkenazicfemale

Easternfemale Total

PLAYER

B

Ashkenazicmale

Easternmale

Ashkenazicfemale

Easternfemale

31

27

22

27

23

28

22

14

15

18

21

18

23

28

23

21

16

17

14

15

14

17

16

13

122

135

118

108

Total 107 87 72 95 62 60 483


fer to Ashkenazic male partners) A two-way variance analysisreveals that this difference is signicant [F(1 256) = 6164P lt 00]13

Our analysis indicates that ethnic discrimination is gender-reliant Section VI will focus on the gender aspect In the interimour discussion will be restricted to games in which both playersare men Figure II illustrates the distribution of money transfers(in the trust game) to Player B depending on his ethnic afliationwhen both players A and B are men

The average transfer by male players to Eastern male play-ers was 562 whereas the average sum that men transferred toAshkenazic male players was 1716 In other words the averagetransfer to Ashkenazic male players was about three times that ofthe amount transferred to Eastern male players The differencebetween the numbers is signicant [F(1 140) = 1013 P lt00)] It is also noteworthy that close to 80 percent of the mentransferred the full NIS20 and achieved the efcient outcomewhen they played with an Ashkenazic male player whereas thisnumber dropped to 12 percent when they played with an Easternmale partner

13 The text reports the ANOVA test results Although this is the standardanalysis some of its assumptions (such as the normality assumption) are prob-lematic with regard to our data For this reason we have also reported in theAppendix the results of the nonparametric Mann-Whitney U-test based on rankWe wish to emphasize that with our data the results of the two tests were similarin all cases

FIGURE ITransfer to Male Players in the Trust Game


Trust among Segments Systematic Discrimination

The results presented in the previous subsections do notnecessarily imply the existence of systematic discriminationagainst a particular ethnic group Previous studies have shownthat even an arbitrary group afliation may affect the way peopletreat others14 Group bias implies that people treat members oftheir own group more favorably than they treat other peopleThus it is possible that discrimination in the trust game merelyreects group bias If players place greater trust in members oftheir own ethnic group and if there were more Ashkenazic stu-dents in the studied population the outcome whereby students ofAshkenazic origin received greater money transfers is notsurprising

In order to check for systematic discrimination against oneparticular ethnic group we asked the students (after they hadplayed the game) to ll out a short questionnaire that included aquestion pertaining to their parentsrsquo country of birth15 Given thisinformation we were able to distinguish between the differentethnic groups of students who participated in the experimentThe population of male students who took on the role of Player A

14 See Tajfel [1982] Tajfel and Turner [1979] Taylor and Moghaddam[1987] and Turner Brown and Tajfel [1979]

15 Providing ethnic afliation is not viewed as an unusual request in manyforms that a typical Israeli student has to ll out throughout hisher years ofstudy

FIGURE IITransfer from Male to MaleAshkenazic and MaleEastern in the Trust Game


was divided into three groups Ashkenazic Eastern and ldquoIsraelirdquoThe latter group also includes students with one parent of East-ern origin and one of Ashkenazic origin Figures IIIa and IIIbsummarize the ndings

Figures IIIa and IIIb show that the pattern of mistrust to-ward male players of Eastern origin is common to all types of(male) players regardless of their ethnic afliation The averagetransfer of Eastern Ashkenazic and Israeli male players to Ash-

FIGURE IIIaTransfer to MaleAshkenazic by Males According to Origin in the Trust Game

FIGURE IIIbTransfer to MaleEastern by Males According to Origin in the Trust Game


kenazic male players was 174 1843 and 161 respectively Wefound no statistically signicant difference between these trans-fers The average transfer to Eastern male players was 528 604and 541 respectively Again we found no statistically signicantdifference between the transfersObservation 2 A systematic mistrust in men of Eastern origin wasfound to be common among men of all ethnic origins In particularthis pattern of mistrust also characterizes men of Eastern originwho discriminate against members of their own group

This paper does not provide any explanation for Observation2 It appears to us however that as part of a cycle of prejudiceand discrimination and as a reaction to their inferior status insociety Eastern Jews have come to believe in these stereotypes ofthemselves

Is Discrimination in the Trust Game Rational

A possible explanation of the discrimination identied in ourexperiment is that people of different ethnic background indeedrespond differently when they play the role of Player B In such acase ethnic discrimination may be rational when based on rele-vant statistical differences between the groups In order to testfor such statistical discrimination in the trust game we examinedthe amounts that Player B transferred back to Player A ClearlyPlayer Brsquos decision regarding the amount he transfers to Player Adepends on the amount transferred to himher in the rst placeThus this comparison can only be established with regard tostudents who received the same amount from Player A Table IIillustrates the average amounts returned by male students ofdifferent origins who received NIS 5 10 15 or 20

In order to statistically compare the distribution of amounts

TABLE IIAVERAGE AMOUNT RETURNED BY MALES ACCORDING TO ETHNIC BACKGROUND AND

THE AMOUNT THEY RECEIVED FROM PLAYER A

Amount given byPlayer A 5 10 15 20

Average returned byAshkenazic male 18 13 172 243

Average returned byEastern male 28 142 167 231


returned by male students of different origins four different com-parisons using the Mann-Whitney U-test were made based on theamount Player A received (NIS 5 10 15 or 20) The hypothesiswhereby the distribution of transfers from Player B back to Player Adoes not depend on Player Brsquos ethnic background cannot be rejectedfor all four comparisons at a 5 level of signicanceObservation 3 We found no evidence for statistical discrimina-tion in the trust game That is we found no evidence that anEastern male player sent back an amount that differed signi-cantly from that returned by an Ashkenazic male player

Although we found no evidence for statistical discriminationthe discrimination that we identied does not necessarily contra-dict rational behavior Players may have misperceptions regard-ing the behavior of players from different ethnic groups and thismisperception may rationally lead to a different strategy choiceWe do not however label such discrimination as ldquostatisticalrdquosince by so doing any discrimination based on stereotypes ratherthan on preferences could be labeled as statistical discrimination

IV TASTE FOR DISCRIMINATION DICTATOR GAME

Ethnic discrimination is not necessarily the result of ethnicstereotypes People may harbor anger dislike or other emotionstoward members of another ethnic group In this case prejudiceis already a part of the playersrsquo emotional makeup The ethnicdiscrimination that we identied in the trust game is thereforenot necessarily a reection of ethnic mistrust ie it could simplybe the result of the above-mentioned taste for discrimination Inorder to distinguish between ethnic stereotype and a taste fordiscrimination we conducted a second experiment in which we letthe students play the dictator game

The dictator game is a two-player game in which at thebeginning of the game Player A gets a xed amount of moneythat he is asked to divide between himself and Player B and thedivision suggested by Player A is carried out In order to renderthis game compatible with the previous trust game we adoptedthe rule whereby any amount transferred from Player A to PlayerB would automatically be tripled by the experimenters Since inthis game Player B is a passive player devoid of any strategicrole any ethnic stereotypes pertaining to potential strategychoices by Player B offer no relevant information Thus if weshould nd that in the dictator game the amounts transferred to


Player B were affected by his ethnic background this would be astrong indication for the existence of a taste for discrimination

The subgame perfect equilibrium of the dictator game is thatPlayer A does not transfer money to Player B However thisequilibrium is based on dening the playersrsquo preferences as maxi-mizing monetary payoffs Numerous experiments however indi-cated that when playing the dictator game players transfer anonnegligible amount to Player B (see Roth [1995] for a survey)This transfer reects the playerrsquos wish for an equitable division ofthe resources provided in the experiment16

Dictator Game Experimental Procedure

The participants in this experiment consisted of 616 Israeliundergraduates At the preliminary stage of the experiment wechose students from the University of Haifa and the Tel AvivAcademic College with typical ethnic names We chose 77 namesfrom each of the four relevant ethnicgender groups

The experiment consisted of one stage in which we ap-proached the students from Tel Aviv University who played therole of Player A and asked them to split the twenty pointsbetween themselves and Player B17 Table III illustrates thenumber of couples matched according to gender and ethnicity

Dictator Game Results

Figure IV illustrates the distribution of transfers (points) byPlayer A (male players at Tel-Aviv University) to Player B (maleplayers at Haifa University) according to the ethnic backgroundof Player B Note that most of the transfers amounted to eitherzero ve or ten points In other words Player A chose either toallocate points equally to allocate money equally or to exploittheir dictator position in order to keep the money for themselvesThe average transfer was 56 to Eastern male players and 51 tomale Ashkenazic players While the average transfers were simi-lar a close examination of the distribution of transfers points toan interesting difference A larger share of the Eastern playersreceived the ve-point transfer whereas a greater portion of theAshkenazic players received either zero or ten-point transfers

In order to test whether the difference between the distribu-

16 Clearly the meaning of ldquoequitable divisionrdquo may vary across people andcultures

17 The instructions given to Player A in this experiment are available fromthe authors upon request


tions is statistically signicant we used the likelihood ratio test(see Davidson and Mackinnon [1993] or the use of the test inStahl [1996] and Camerer and Ho [1999]) We used as modes ofbehavior the choice of 0 5 10 or random choice We took themaximum likelihood parameters of population 1 (Ashkenazicmales) and imposed them on population 2 (Eastern males) Theresulting likelihood is the restricted model likelihood In contrastthe maximum likelihood of population 2 without the imposition ofparameters is the unrestricted likelihood Twice the difference

TABLE IIIPARTICIPATING PAIRS IN THE DICTATOR GAME ACCORDING TO GENDER

AND ETHNICITY

Player A

Israelimale

Ashkenazicmale

Easternmale

Israelifemale

Ashkenazicfemale

Easternfemale Total

PLAYER

B

Ashkenazicmale

Easternmale

Ashkenazicfemale

Easternfemale

22

24

15

22

12

14

16

13

10

9

10

13

17

14

19

13

11

9

8

10

5

7

9

6

77

77

77

77

Total 83 55 42 63 38 27 308

FIGURE IVTransfer by Males to Males According to Origin in the Dictator Game


between the unrestricted log-likelihood and the restricted log-likelihood corresponds to the likelihood ratio static The re-stricted log-likelihood of population 2 is 64 and the unrestrictedis 592 Twice the difference is 96 which given a x 2 distributionwith three degrees of freedom would have a p-value of 0022 Inother words the difference between the populations is marginallysignicant

A comparison of the transfers of players from different ethnicgroups indicates that the distributions were similar18 We thusconclude that there was no evidence for the group effect Playersin our dictator game experiment did not behave favorably towardplayers of their own groupObservation 4 The results of the dictator game indicate thatthere is some differential treatment of groups by ethnicity but noclear systematic taste for discrimination The distribution oftransfers to the Ashkenazic players was only marginally differentfrom the distribution of transfers to Eastern players The averagetransfers were similar

While we found some evidence of a taste for discriminationthese types of preferences cannot explain the discrimination weidentied in the trust game Since in the dictator game theaverage transfer to Eastern and Ashkenazic players was similarsuch taste for discrimination cannot explain the huge differenceof transfers in the trust game Thus we conclude that discrimi-nation in the trust game is due to ethnic stereotyping that affectsthe playersrsquo beliefs with regard to Player Brsquos potential reaction

V STEREOTYPES REACTION TO UNFAIR TREATMENT

By changing the game again we were able to trace othertypes of stereotypes A number of Israelis have often argued thatpeople of Eastern ethnic background tend to be more driven by asense of ldquohonorrdquo or to react harshly if treated unfairly People whoopenly expressed similar stereotypes in Israel were denounced asbigots and frequently punished by society The question iswhether such beliefs are actually held by a small minority inIsraeli society or whether they are much more widespread than

18 These distributions of transfers were common among both ethnic groupsIn our experiment both Ashkenazic and Eastern male players transferred onaverage 5 points to Ashkenazic players whereas Israeli male players transferred52 on the average The average transfers to Eastern players were 58 52 and 58by Israeli Ashkenazic and Eastern male players respectively


people would care to admit In order to answer this question wechanged the game again and studied the ldquoultimatum gamerdquo Theultimatum game is a simple two-player two-stage game In therst stage Player A is given a xed amount of money and thenasked to suggest a division of this amount between himself andPlayer B At the second stage it is Player Brsquos turn to either acceptor reject the offer If the offer is accepted the players get theirsuggested shares but if it is rejected neither player gets any-thing As in the previous two games any amount transferred toPlayer B is automatically tripled by the experimenters

The subgame perfect equilibrium of the ultimatum game issuch that Player A makes a zero (or minimal positive) offer toPlayer B while Player B accepts any division suggested to himThis equilibrium is based on the standard assumption that indi-viduals strive to maximize their monetary payoffs The ultima-tum game has been extensively discussed in the literature Nu-merous experiments suggest that Player A typically transfers anonnegligible amount while Player B occasionally rejects lowoffers19 Player Arsquos nonnegligible offers can be partially explainedby his wish to achieve an equitable outcome but also by his beliefthat Player B may reject a low ldquounfairrdquo offer Since Player B hasa strategic role in the ultimatum game ethnic stereotypes mayaffect Player Arsquos beliefs with regard to Player Brsquos potentialresponse

Ultimatum Game Experimental Design

The ultimatum gamersquos experimental design is similar to thedesign of the trust and dictator games Participants in this ex-periment consisted of 680 Israeli undergraduates At the pre-liminary stage of the experiment we chose students from theUniversity of Haifa and the Tel Aviv Academic College withtypical ethnic names

The experiment consisted of two stages At the rst stage thestudents from Tel Aviv University (Player A) received twentypoints and were asked to propose a division between themselvesand Player B At the second stage we approached the students atthe University of Haifa and the Tel Aviv Academic College whoplayed the role of Player B and asked them to decide whether to

19 See the original experiment by Guth Schmittberger and Schwarze[1982] and the surveys in Camerer and Thaler [1995] Guth [1995] and Roth[1995]


accept the proposed division We then paid the students accord-ingly20 Table IV illustrates the couples that were matched ac-cording to gender and ethnicity

Ultimatum Game Results

Figure V illustrates the distribution of (points) transferred inthe ultimatum game experiment Again we only considered trans-fers by male players to male players It is interesting to note thatalmost 90 percent of the transfers are either ve or ten pointsPlayers chose either an equal allocation of points or an equalallocation of money However surprisingly these choices de-pended very much on the game partnersrsquo ethnic afliation East-ern male players received on average 84 while the averagetransfer to Ashkenazic male players was only 59 By using thelikelihood ratio test we found that the restricted log-likelihood ofpopulation 2 was 169 and the unrestricted 53 Twice the differ-ence is 232 which given a x 2 distribution with 3 degrees offreedom would have a p-value of less than 0001 Hence weconclude that the difference is highly signicantObservation 5 There is ethnic discrimination in the ultimatumgame Eastern players receive larger transfers than Ashkenazic

20 The instructions for this experiment are available from the authors uponrequest

TABLE IVPARTICIPATING PAIRS IN THE ULTIMATUM GAME ACCORDING TO GENDER AND

ETHNICITY

Player A

Israelimale

Ashkenazicmale

Easternmale

Israelifemale

Ashkenazicfemale

Easternfemale Total

PLAYER

B

Ashkenazicmale

Easternmale

Ashkenazicfemale

Easternfemale

23

22

15

19

10

14

10

12

14

19

13

9

26

18

13

14

17

16

10

8

10

11

9

8

100

100

70

70

Total 79 46 55 71 51 38 340


players This discrimination is probably the outcome of a commonethnic stereotype in Israeli society according to which men ofEastern origin are believed to react more harshly if treatedunfairly

Is discrimination in the ultimatum game rational In order totest for statistical differences in the reaction of individuals weexamined the rejection rates of Player B when being offered 5points Seven (10 percent) out of the 67 Ashkenazic males whowere offered 5 points rejected them whereas 3 (8 percent) out ofthe 38 Eastern males did not We conclude that there is noevidence for statistical discrimination in the ultimatum game

VI GENDER AND DISCRIMINATION

When considering the relationship between gender and eth-nic discrimination there are two separate but related questions(i) Is there discrimination against women and does it depend onethnic afliation (ii) Do women themselves discriminate be-tween Ashkenazic and Eastern players While gender may affectthe outcome of all three experiments discussed in this paper wehave limited the discussion in this section to the trust game only

GenderEthnic Discrimination

We shall now discuss the results of the trust game experi-ment in which women played the role of Player B Our rstobservation in this experiment is that the average transfer to

FIGURE VTransfer by Males to Males According to Origin in the Ultimatum Game


female players was 1063 which is similar to the average transferto male players namely 1142 Thus at rst glance there is noevidence of gender-based discrimination This conclusion how-ever is misleading as we need to examine the genderethniccombination

Figure VI describes the distribution of transfers to femaleplayers based on their ethnic background The average transfer toAshkenazic female players is 1102 while the average transfer toEastern female players is 1041 The difference between these twodistributions is statistically insignicant [F(1 222) = 352P gt 05)]Observation 6

(i) We found no evidence of ethnic discrimination betweenwomen in the trust game

(ii) The average transfer from male players to Ashkenazicwomen was 1102 whereas the average transfer to Ash-kenazic men was 1515 A comparison between the twodistributions implies that Ashkenazic women are lesstrusted than Ashkenazic men are The difference ishighly signicant [F(1 238) = 1978 P lt 00)] On theother hand the average transfer to Eastern women was1041 while the average transfer to Eastern men was806 A comparison between the two distributions implies

FIGURE VITransfer to Female Players in the Trust Game


that Eastern women are more trusted than Eastern menare [F(1 240) = 557 P lt 019)]

It is noteworthy that the above observation is consistentwith several early studies on discrimination of Afro-Americanwomen in the United States For example Epstein [1993]pointed out that although one may expect that Afro-Americanwomen may face more severe discrimination (double subordi-nation) this may actually result in a relative advantage Offurther interesting note is that the ethnic earning gaps inIsrael among women are actually smaller than those amongmen (see Cohen and Haberfeld [1998]) See also Blau Ferberand Winkler [1997] who nd that in the United States thedifferences in earnings between Afro-Americans and whitesare smaller for females than for males

Do Women and Men Discriminate in the Same Way

After discussing behavior toward women we now addresswomenrsquos behavior namely the way women play the trust gamewhen they are assigned as Player A Once again we have onlyconsidered the case in which Player B is a male player (Note thatwe did not nd any evidence of discrimination between women ofdifferent ethnic backgrounds)

Figure VII illustrates the distribution of transfers by female

FIGURE VIITransfer from Females to MaleAshkenazic and MaleEastern in

the Trust Game


players to Ashkenazic and Eastern male players The comparisonof Figure VII (the distribution of transfers by female players) toFigure II (the distribution of transfers by male players) is strik-ing men responded strongly to ethnic stereotypes whereaswomen did not The average transfer by women to Eastern maleplayers was 1094 whereas the average transfer by women toAshkenazic male players was 1253 The difference between thedistributions is insignicant [F(1 113) = 148 P gt 05)] Bycontrast the average transfer made by male players to Easternmale players was 562 whereas the average amount that mentransferred to Ashkenazic male players was 1716

When women played the trust game with other women theaverage transfer to Ashkenazic women was 113 and the averagetransfer to Eastern women was 1038 A comparison of theseoutcomes with the above results implies the followingObservation 7

(i) We found no evidence that womenrsquos trust in theirgame partners is based on ethnic afliation or on gen-der21

(ii) Men trust Ashkenazic male players more than women doOn the other hand men trust Eastern male players lessthan women do

Our nding is consistent with the observation that the ethnicearning gaps in Israel among women are indeed smaller thanamong men and that in recent years Ashkenazic men have in-creased the earning gaps between them and other groups (seeCohen and Haberfeld [1998])

VII CONCLUDING REMARKS

Ethnic stereotyping may result in different types of discrim-ination in schooling wages the product market etc An impor-tant difference between the different types of discrimination istheir degree of observability An important aspect of discrimina-tion pertains to interpersonal relations (such as trust and will-ingness to cooperate) which are difcult to observe Wages on

21 It was common knowledge in our experiment that all players were stu-dents It is possible that the ethnical pattern of trust and mistrust also dependson the game partnersrsquo level of education Thus our experiments indicate thatwomen do not discriminate based on ethnic background when they play witheducated partners We cannot however reach similar conclusions for the entirepopulation of potential partners


the other hand are relatively observable and have therefore beenthe focus of most of the studies on discrimination If we were todevise public policies aimed at lessening ethnic discrimination itwould probably be easier to implement such policies with observ-able cases of discrimination We may have antidiscriminationlaws claiming that wages and school admissions must be inde-pendent of ethnic background or gender but legislation withrespect to trust and interpersonal relationships are ineffective

This paper focused on how to examine the types ofdiscrimination that are not easily observable By using gameexperiments one can identify these types of discrimination sincethe experiments are based on preferences revealed by individualswhen facing a strategic situation with another individualClearly such experiments cannot be used as a legislative toolhowever in our opinion they could serve as an important educa-tional tool To begin with these experiments could be used toconfront individuals with their own behavior In a class in whichthis experiment was used as a demonstration participants wereinformed of its true purpose at the end of the experiment Theyreacted by saying that they had considered neither the name northe ethnic signal as relevant information and that their ownprediction was that the distribution of transfers to Eastern andAshkenazic players would be similar After the brief classroomdebate the experimenter opened the forms and showed the stu-dents the distribution of transfers in the class The reaction wasone of silence and total amazement There was no disagreementin the class with regard to the results since they were overwhelm-ing However the silence was mainly due to the recognition thatindeed even without being aware of it the students in the classhad blatantly discriminated against one group This is preciselythe advantage of using experiments rather than questionnaireswhich are commonly used in many of the sociological discrimina-tion surveys The experiments illustrate peoplersquos behavior ratherthan what people believe to be their own behavior

APPENDIX 1 INSTRUCTIONS FOR PLAYER A(TRANSLATED FROM HEBREW)

Welcome to this experiment in decision-making during whichyou may earn some money that will be paid to you privately andin cash at the end

The interaction in the experiment will be in pairs You are


called Student A and the student you are matched with is calledStudent B The participants in the role of Student B are from theUniversity of Haifa

At the beginning of the experiment you will receive NIS 20and Student B will not receive any money You are asked todecide whether you wish to transfer any amount of the NIS 20 tothe student you are matched with and if so how much We willtriple the amount you transfer and give it to Student B that isfor every NIS 1 that you transfer Student B will receive NIS 3

In a few days time we will ask Student B to decide if (s)hewants to return any of the money (s)he received (three times whatyou sent) and if so how much This amount will not be tripledThis will conclude the experiment and the money will be paidName of the student you are matched with (Student B)Your nameAmount of money you wish to transfer to Student B (Pleaseremember that this amount should be between NIS 0 and NIS 20)

APPENDIX 2 RESULTS OF THE MANN-WHITNEY U-TEST

BASED ON RANKS

As argued in the text although ANOVA is the standard testin such studies some of the testsrsquo assumptions most notably thenormality assumption are not fullled by our data Therefore inorder to support the statistical analysis we have also conducteda nonparametric test As it turns out the two tests result insimilar conclusions based on our data The results of the nonpara-metric test are provided below in an order corresponding to thatof the ANOVA test reported in the paper

APPENDIX 3 RESULTS OF THE NONPARAMETRIC TEST

Comparison of Z Plt

Transfer to Ashkenazic malesEastern males 7363 000Transfer to malesfemales 1390 164Transfer to Ashkenazic femalesEastern females 682 527Transfer to Ashkenazic malesAshkenazic females 4672 000Transfer to Eastern malesEastern females 2073 038Transfer from females to Ashkenazic malesEastern males 1288 198Transfer from males to Ashkenazic malesEastern males 7904 000

THE EITAN BERGLAS SCHOOL OF ECONOMICS TEL AVIV UNIVERSITY

FACULTY OF INDUSTRIAL ENGINEERING AND MANAGEMENT TECHNION


REFERENCES

Altonji Joseph and Rebecca Blank ldquoRace and Gender in the Labor Marketrdquo inOrley Ashenfelter and David Card eds Handbook of Labor EconomicsVolume 3C (Amsterdam Elsevier Science North-Holland 1999)

Amir Shmuel ldquoTrends in Earnings Gaps among Jewish Men during the 70rsquos byCountry of Originrdquo Survey No 63 Bank of Israel 1987 43ndash 64

Andreoni James and Lise Vesterlund ldquoWhich Is the Fair Sex Gender Differ-ences in Altruismrdquo Quarterly Journal of Economics CXVI (2001) 293ndash312

Arrow Kenneth ldquoGifts and Exchangesrdquo Philosophy and Public Affairs I (1972)343ndash362

mdashmdash ldquoThe Theory of Discriminationrdquo in Orley Ashenfelter and Albert Rees edsDiscrimination in Labor Markets (Princeton NJ Princeton University Press1973) pp 3ndash33

mdashmdash ldquoWhat Has Economics to Say about Racial Discriminationrdquo Journal ofEconomic Perspectives XII (1998) 91ndash100

Becker Gary The Economics of Discrimination (Chicago University of ChicagoPress 1957)

mdashmdash ldquoThe Evidence against Banks Doesnrsquot Prove Biasrdquo Business Week April 191993

Berg Joyce John Dickhaut and Kevin McCabe ldquoTrust Reciprocity and SocialHistoryrdquo Games and Economic Behavior X (1995) 122ndash142

Blau Francine Marianne Ferber and Anne Winkler Economics of Women Menand Work (Englewood Cliffs NJ Prentice Hall Business Publishing 1998)

Bolton Gary and Elena Katok ldquoAn Experimental Test for Gender Differences inBenecent Behaviorrdquo Economics Letters XLVIII (1995) 287ndash292

Camerer Colin and Teck-Hua Ho ldquoExperience-Weighted Attraction Learning inNormal Form Gamesrdquo Econometrica LXVII (1999) 827ndash874

Camerer Colin and Richard Thaler ldquoUltimatums Dictators and Mannersrdquo Jour-nal of Economic Perspectives IX (1995) 209 ndash219

Camerer Colin and Keith Weigelt ldquoExperimental Tests of a Sequential Equilib-rium Reputation Modelrdquo Econometrica LVI (1988) 1ndash36

Cohen Yinon and Yitchak Haberfeld ldquoSecond-Generation Jewish Immigration inIsrael Have the Ethnic Gaps in Schooling and Earnings Declinedrdquo Ethnicand Racial Studies XXI (1998) 507ndash528

Croson Rachel and Nancy Buchan ldquoGender and Culture International Experi-mental Evidence from Trust Gamesrdquo American Economic Review Papers andProceedings LXXXIX (1999) 386ndash391

Darity William and Patrick Mason ldquoEvidence on Discrimination in Employ-ment Codes of Color Codes of Genderrdquo Journal of Economic Perspectives XII(1998) 63ndash90

Davidson Russell and James Mackinnon Estimation and Inference Econometrics(Oxford Oxford University Press 1993)

Eckel Catherine and Philip Grossman ldquoChivalry and Solidarity in UltimatumGamesrdquo Virginia Polytechnic Institute and State University Department ofEconomics Working Paper E92-23 1997

Eckel Catherine and Philip Grossman ldquoAre Women Less Selsh than MenEvidence from Dictator Experimentsrdquo Economic Journal CVIII (1998) 726 ndash735

Eisenstadt Shmuel The Transformation of Israeli Society (London Weidenfeldand Nicolson 1985)

Epstein Cynthia ldquoPositive Effects of the Multiple Negative Explaining theSuccess of Black Professional Womenrdquo American Journal of SociologyLXXVIII (1993) 150 ndash173

Fehr Ernst Simon Gachter and Georg Kirchsteiger ldquoReciprocity as a ContractEnforcement Device Experimental Evidencerdquo Econometrica LXV (1997)833ndash860

Fehr Ernst Georg Kirchsteiger and Arno Riedl ldquoDoes Fairness Prevent MarketClearing An Experimental Investigationrdquo Quarterly Journal of EconomicsCVIII (1993) 437ndash460

Fershtman Chaim and Uri Gneezy ldquoSecular-Religious Relationship in Israelrdquo(Hebrew) mimeo Tel Aviv University 2000


Fix Michael and Raymond Struyk eds Clear and Convincing Evidence Mea-surement of Discrimination in America (Washington DC Urban InstitutePress 1993)

Guth Werner ldquoOn Ultimatum Bargaining Experiments A Personal ReviewrdquoJournal of Economic Behavior and Organization XXVII (1995) 329ndash344

Guth Werner P Ockenfels and M Wendel ldquoEfciency by Trust in FairnessMultiperiod Ultimatum Bargaining Experiments with an Increasing CakerdquoInternational Journal of Game Theory XXII (1994) 51ndash77

Guth Werner R Schmittberger and B Schwarze ldquoAn Experimental Analysis ofUltimatum Bargainingrdquo Journal of Economic Behavior and Organization III(1982) 367ndash388

Haberfeld Yitchak ldquoImmigration and Ethnic Origin the Effect of DemographicAttributes on Earnings of Israeli Men and Womenrdquo International MigrationReview XXVII (1992) 286 ndash305

Heckman James ldquoDetecting Discriminationrdquo Journal of Economic PerspectivesXII (1998) 101ndash116

Holm Hakan ldquoGender Based Focal Pointsrdquo Games and Economic Behavior 2000forthcoming

Knack Stephen and Philip Keefer ldquoDoes Social Capital Have an EconomicPayoff A Cross-Country Investigationrdquo Quarterly Journal of EconomicsCXII (1997) 1251ndash1288

Ladd Helen ldquoEvidence on Discrimination in Mortgage Lendingrdquo Journal ofEconomic Perspectives XII (1998) 41ndash62

Loury Glenn ldquoDiscrimination in the Post-Civil Rights Era Beyond Market In-teractionsrdquo Journal of Economic Perspectives XII (1998) 117ndash126

Mark Nili ldquoEthnic Gaps in Earning and Consumption in Israelrdquo EconomicQuarterly XLI (1994) 55ndash77

McKelvey Richard and Thomas Palfrey ldquoAn Experimental-Study of the Centi-pede Gamerdquo Econometrica LX (1992) 803ndash 836

Neumark David ldquoSex Discrimination in Restaurant Hiring An Audit StudyrdquoQuarterly Journal of Economics CXI (1996) 915ndash941

Roth Alvin ldquoBargaining Experimentsrdquo in J Kagel and A Roth eds The Hand-book of Experimental Economics (Princeton Princeton University Press1995)

Semyonov Moshe and Vered Kraus ldquoGender Ethnicity and Earnings Inequal-ity The Israeli Experiencerdquo International Journal of Comparative SociologyXXIV (1983) 258ndash72

Semyonov Moshe and Tamar Lerenthal ldquoCountry of Origin Gender and theAttainment of Economic Status A Study of Stratication in the JewishPopulation in Israelrdquo Research in Social Stratication and Mobility X (1991)325ndash343

Shavit Yossi ldquoTracking and Ethnicity in Israeli Secondary Educationrdquo AmericanSociological Review XLIX (1984) 210 ndash220

Stahl Dale ldquoBoundedly Rational Rule Learning in a Guessing Gamerdquo Games andEconomic Behavior XVI (1996) 303ndash330

Tajfel Henri ldquoSocial Psychology of Intergroup Relationsrdquo Annual Review ofPsychology XXXIII (1982) 1ndash39

Tajfel Henri and John Turner ldquoAn Integrative Theory of Intergroup Conictrdquo inW G Austin and S Worchel eds The Social Psychology of IntergroupRelations (Monterey CA BrooksCole 1979)

Taylor Donald and Fathali Moghaddam Theories of Intergroup Relations Inter-national Social Psychological Perspectives (New York Praeger 1987)

Turner John Rupert Brown and Henri Tajfel ldquoSocial Comparison and GroupInterest in Intergroup Favoritismrdquo European Journal of Social PsychologyIX (1979) 187ndash204

Yinger John ldquoEvidence on Discrimination in Consumer Marketsrdquo Journal ofEconomic Perspectives XII (1998) 23ndash40


has been extensively discussed in the literature2 However whilethese studies are useful in providing evidence for discriminationin different markets they are less so when it comes to explainingthe type of discrimination identied In this paper we propose anexperimental approach to studying the different aspects of dis-crimination In these experiments individuals played a variety ofgames against players of different ethnic groups We subse-quently examined how the ethnic afliation of game partnersaffected the strategy chosen by the players The outcome of theseexperiments may clarify the following questions

(i) Is there discrimination (differential treatment) based onethnic afliation

(ii) Does the discrimination reect a group bias in that eachplayer favors players of his own group or is there asystematic discrimination against one or several ethnicgroups

(iii) Is this discrimination based on a taste for discrimina-tion or is it the outcome of ethnic stereotyping thataffects the playersrsquo assessment regarding their gamepartnersrsquo strategic responses or relevant characteristics

(iv) Are the ethnic stereotypes accurateThis experimental approach was used to examine ethnic

discrimination within Israeli Jewish society3 The latter is char-acterized by an ethnic structure based primarily on country oforigin The two major ethnic groups are Ashkenazic Jews (Euro-pean and American immigrants and their Israeli-born offspring)and Eastern Jews (Asian and African immigrants and their Is-raeli-born offspring) There are persistent economic gaps betweenthe two ethnic segments Ashkenazic immigrants achieve higherlevels of education and earnings than do Eastern immigrantsand these gaps continue to prevail among second-generation im-migrants (see Cohen and Haberfeld [1998] Eisenstadt [1985]

2 There is ample literature on audit studies showing direct evidence ofdiscrimination in employment housing and credit (see for example Fix andStruyk [1993] as well as discussion and references on the subject in Altonji andBlank [1999] and Neumark [1996]) See also the recent special Symposium issueon Discrimination in Product Credit and Labor Markets that appeared in theJournal of Economic Perspectives [1998] and the following papers in this sympo-sium Yinger [1998] Ladd [1998] Darity and Mason [1998] Arrow [1998] Heck-man [1998] and Loury [1998]

3 A clearly signicant and heated segmentation in Israeli society is thatbetween Arabs and Jews but we reserve this issue for future research Anothermajor segmentation in Israeli Jewish society is the one between secular andOrthodox Jews






















GROUPS IN ISRAEL






















Segmented Society



Player A

Israelimale

Ashkenazicmale

Easternmale

Israelifemale

Ashkenazicfemale

Easternfemale Total

PLAYER

B

Ashkenazicmale

Easternmale

Ashkenazicfemale

Easternfemale

31

27

22

27

23

28

22

14

15

18

21

18

23

28

23

21

16

17

14

15

14

17

16

13

122

135

118

108

Total 107 87 72 95 62 60 483


















































AND ETHNICITY

Player A

Israelimale

Ashkenazicmale

Easternmale

Israelifemale

Ashkenazicfemale

Easternfemale Total

PLAYER

B

Ashkenazicmale

Easternmale

Ashkenazicfemale

Easternfemale

22

24

15

22

12

14

16

13

10

9

10

13

17

14

19

13

11

9

8

10

5

7

9

6

77

77

77

77

Total 83 55 42 63 38 27 308






















ETHNICITY

Player A

Israelimale

Ashkenazicmale

Easternmale

Israelifemale

Ashkenazicfemale

Easternfemale Total

PLAYER

B

Ashkenazicmale

Easternmale

Ashkenazicfemale

Easternfemale

23

22

15

19

10

14

10

12

14

19

13

9

26

18

13

14

17

16

10

8

10

11

9

8

100

100

70

70

Total 79 46 55 71 51 38 340






















the Trust Game





















BASED ON RANKS



Comparison of Z Plt





REFERENCES







































































GROUPS IN ISRAEL






















Segmented Society



Player A

Israelimale

Ashkenazicmale

Easternmale

Israelifemale

Ashkenazicfemale

Easternfemale Total

PLAYER

B

Ashkenazicmale

Easternmale

Ashkenazicfemale

Easternfemale

31

27

22

27

23

28

22

14

15

18

21

18

23

28

23

21

16

17

14

15

14

17

16

13

122

135

118

108

Total 107 87 72 95 62 60 483


















































AND ETHNICITY

Player A

Israelimale

Ashkenazicmale

Easternmale

Israelifemale

Ashkenazicfemale

Easternfemale Total

PLAYER

B

Ashkenazicmale

Easternmale

Ashkenazicfemale

Easternfemale

22

24

15

22

12

14

16

13

10

9

10

13

17

14

19

13

11

9

8

10

5

7

9

6

77

77

77

77

Total 83 55 42 63 38 27 308






















ETHNICITY

Player A

Israelimale

Ashkenazicmale

Easternmale

Israelifemale

Ashkenazicfemale

Easternfemale Total

PLAYER

B

Ashkenazicmale

Easternmale

Ashkenazicfemale

Easternfemale

23

22

15

19

10

14

10

12

14

19

13

9

26

18

13

14

17

16

10

8

10

11

9

8

100

100

70

70

Total 79 46 55 71 51 38 340






















the Trust Game





















BASED ON RANKS



Comparison of Z Plt





REFERENCES
































































GROUPS IN ISRAEL






















Segmented Society



Player A

Israelimale

Ashkenazicmale

Easternmale

Israelifemale

Ashkenazicfemale

Easternfemale Total

PLAYER

B

Ashkenazicmale

Easternmale

Ashkenazicfemale

Easternfemale

31

27

22

27

23

28

22

14

15

18

21

18

23

28

23

21

16

17

14

15

14

17

16

13

122

135

118

108

Total 107 87 72 95 62 60 483


















































AND ETHNICITY

Player A

Israelimale

Ashkenazicmale

Easternmale

Israelifemale

Ashkenazicfemale

Easternfemale Total

PLAYER

B

Ashkenazicmale

Easternmale

Ashkenazicfemale

Easternfemale

22

24

15

22

12

14

16

13

10

9

10

13

17

14

19

13

11

9

8

10

5

7

9

6

77

77

77

77

Total 83 55 42 63 38 27 308






















ETHNICITY

Player A

Israelimale

Ashkenazicmale

Easternmale

Israelifemale

Ashkenazicfemale

Easternfemale Total

PLAYER

B

Ashkenazicmale

Easternmale

Ashkenazicfemale

Easternfemale

23

22

15

19

10

14

10

12

14

19

13

9

26

18

13

14

17

16

10

8

10

11

9

8

100

100

70

70

Total 79 46 55 71 51 38 340






















the Trust Game





















BASED ON RANKS



Comparison of Z Plt





REFERENCES



























































GROUPS IN ISRAEL






















Segmented Society



Player A

Israelimale

Ashkenazicmale

Easternmale

Israelifemale

Ashkenazicfemale

Easternfemale Total

PLAYER

B

Ashkenazicmale

Easternmale

Ashkenazicfemale

Easternfemale

31

27

22

27

23

28

22

14

15

18

21

18

23

28

23

21

16

17

14

15

14

17

16

13

122

135

118

108

Total 107 87 72 95 62 60 483


















































AND ETHNICITY

Player A

Israelimale

Ashkenazicmale

Easternmale

Israelifemale

Ashkenazicfemale

Easternfemale Total

PLAYER

B

Ashkenazicmale

Easternmale

Ashkenazicfemale

Easternfemale

22

24

15

22

12

14

16

13

10

9

10

13

17

14

19

13

11

9

8

10

5

7

9

6

77

77

77

77

Total 83 55 42 63 38 27 308






















ETHNICITY

Player A

Israelimale

Ashkenazicmale

Easternmale

Israelifemale

Ashkenazicfemale

Easternfemale Total

PLAYER

B

Ashkenazicmale

Easternmale

Ashkenazicfemale

Easternfemale

23

22

15

19

10

14

10

12

14

19

13

9

26

18

13

14

17

16

10

8

10

11

9

8

100

100

70

70

Total 79 46 55 71 51 38 340






















the Trust Game





















BASED ON RANKS



Comparison of Z Plt





REFERENCES





















































GROUPS IN ISRAEL






















Segmented Society



Player A

Israelimale

Ashkenazicmale

Easternmale

Israelifemale

Ashkenazicfemale

Easternfemale Total

PLAYER

B

Ashkenazicmale

Easternmale

Ashkenazicfemale

Easternfemale

31

27

22

27

23

28

22

14

15

18

21

18

23

28

23

21

16

17

14

15

14

17

16

13

122

135

118

108

Total 107 87 72 95 62 60 483


















































AND ETHNICITY

Player A

Israelimale

Ashkenazicmale

Easternmale

Israelifemale

Ashkenazicfemale

Easternfemale Total

PLAYER

B

Ashkenazicmale

Easternmale

Ashkenazicfemale

Easternfemale

22

24

15

22

12

14

16

13

10

9

10

13

17

14

19

13

11

9

8

10

5

7

9

6

77

77

77

77

Total 83 55 42 63 38 27 308






















ETHNICITY

Player A

Israelimale

Ashkenazicmale

Easternmale

Israelifemale

Ashkenazicfemale

Easternfemale Total

PLAYER

B

Ashkenazicmale

Easternmale

Ashkenazicfemale

Easternfemale

23

22

15

19

10

14

10

12

14

19

13

9

26

18

13

14

17

16

10

8

10

11

9

8

100

100

70

70

Total 79 46 55 71 51 38 340






















the Trust Game





















BASED ON RANKS



Comparison of Z Plt





REFERENCES




































































Segmented Society



Player A

Israelimale

Ashkenazicmale

Easternmale

Israelifemale

Ashkenazicfemale

Easternfemale Total

PLAYER

B

Ashkenazicmale

Easternmale

Ashkenazicfemale

Easternfemale

31

27

22

27

23

28

22

14

15

18

21

18

23

28

23

21

16

17

14

15

14

17

16

13

122

135

118

108

Total 107 87 72 95 62 60 483


















































AND ETHNICITY

Player A

Israelimale

Ashkenazicmale

Easternmale

Israelifemale

Ashkenazicfemale

Easternfemale Total

PLAYER

B

Ashkenazicmale

Easternmale

Ashkenazicfemale

Easternfemale

22

24

15

22

12

14

16

13

10

9

10

13

17

14

19

13

11

9

8

10

5

7

9

6

77

77

77

77

Total 83 55 42 63 38 27 308






















ETHNICITY

Player A

Israelimale

Ashkenazicmale

Easternmale

Israelifemale

Ashkenazicfemale

Easternfemale Total

PLAYER

B

Ashkenazicmale

Easternmale

Ashkenazicfemale

Easternfemale

23

22

15

19

10

14

10

12

14

19

13

9

26

18

13

14

17

16

10

8

10

11

9

8

100

100

70

70

Total 79 46 55 71 51 38 340






















the Trust Game





















BASED ON RANKS



Comparison of Z Plt





REFERENCES



























































Segmented Society



Player A

Israelimale

Ashkenazicmale

Easternmale

Israelifemale

Ashkenazicfemale

Easternfemale Total

PLAYER

B

Ashkenazicmale

Easternmale

Ashkenazicfemale

Easternfemale

31

27

22

27

23

28

22

14

15

18

21

18

23

28

23

21

16

17

14

15

14

17

16

13

122

135

118

108

Total 107 87 72 95 62 60 483


















































AND ETHNICITY

Player A

Israelimale

Ashkenazicmale

Easternmale

Israelifemale

Ashkenazicfemale

Easternfemale Total

PLAYER

B

Ashkenazicmale

Easternmale

Ashkenazicfemale

Easternfemale

22

24

15

22

12

14

16

13

10

9

10

13

17

14

19

13

11

9

8

10

5

7

9

6

77

77

77

77

Total 83 55 42 63 38 27 308






















ETHNICITY

Player A

Israelimale

Ashkenazicmale

Easternmale

Israelifemale

Ashkenazicfemale

Easternfemale Total

PLAYER

B

Ashkenazicmale

Easternmale

Ashkenazicfemale

Easternfemale

23

22

15

19

10

14

10

12

14

19

13

9

26

18

13

14

17

16

10

8

10

11

9

8

100

100

70

70

Total 79 46 55 71 51 38 340






















the Trust Game





















BASED ON RANKS



Comparison of Z Plt





REFERENCES




















































Segmented Society



Player A

Israelimale

Ashkenazicmale

Easternmale

Israelifemale

Ashkenazicfemale

Easternfemale Total

PLAYER

B

Ashkenazicmale

Easternmale

Ashkenazicfemale

Easternfemale

31

27

22

27

23

28

22

14

15

18

21

18

23

28

23

21

16

17

14

15

14

17

16

13

122

135

118

108

Total 107 87 72 95 62 60 483


















































AND ETHNICITY

Player A

Israelimale

Ashkenazicmale

Easternmale

Israelifemale

Ashkenazicfemale

Easternfemale Total

PLAYER

B

Ashkenazicmale

Easternmale

Ashkenazicfemale

Easternfemale

22

24

15

22

12

14

16

13

10

9

10

13

17

14

19

13

11

9

8

10

5

7

9

6

77

77

77

77

Total 83 55 42 63 38 27 308






















ETHNICITY

Player A

Israelimale

Ashkenazicmale

Easternmale

Israelifemale

Ashkenazicfemale

Easternfemale Total

PLAYER

B

Ashkenazicmale

Easternmale

Ashkenazicfemale

Easternfemale

23

22

15

19

10

14

10

12

14

19

13

9

26

18

13

14

17

16

10

8

10

11

9

8

100

100

70

70

Total 79 46 55 71 51 38 340






















the Trust Game





















BASED ON RANKS



Comparison of Z Plt





REFERENCES



































































































AND ETHNICITY

Player A

Israelimale

Ashkenazicmale

Easternmale

Israelifemale

Ashkenazicfemale

Easternfemale Total

PLAYER

B

Ashkenazicmale

Easternmale

Ashkenazicfemale

Easternfemale

22

24

15

22

12

14

16

13

10

9

10

13

17

14

19

13

11

9

8

10

5

7

9

6

77

77

77

77

Total 83 55 42 63 38 27 308






















ETHNICITY

Player A

Israelimale

Ashkenazicmale

Easternmale

Israelifemale

Ashkenazicfemale

Easternfemale Total

PLAYER

B

Ashkenazicmale

Easternmale

Ashkenazicfemale

Easternfemale

23

22

15

19

10

14

10

12

14

19

13

9

26

18

13

14

17

16

10

8

10

11

9

8

100

100

70

70

Total 79 46 55 71 51 38 340






















the Trust Game





















BASED ON RANKS



Comparison of Z Plt





REFERENCES





























































































AND ETHNICITY

Player A

Israelimale

Ashkenazicmale

Easternmale

Israelifemale

Ashkenazicfemale

Easternfemale Total

PLAYER

B

Ashkenazicmale

Easternmale

Ashkenazicfemale

Easternfemale

22

24

15

22

12

14

16

13

10

9

10

13

17

14

19

13

11

9

8

10

5

7

9

6

77

77

77

77

Total 83 55 42 63 38 27 308






















ETHNICITY

Player A

Israelimale

Ashkenazicmale

Easternmale

Israelifemale

Ashkenazicfemale

Easternfemale Total

PLAYER

B

Ashkenazicmale

Easternmale

Ashkenazicfemale

Easternfemale

23

22

15

19

10

14

10

12

14

19

13

9

26

18

13

14

17

16

10

8

10

11

9

8

100

100

70

70

Total 79 46 55 71 51 38 340






















the Trust Game





















BASED ON RANKS



Comparison of Z Plt





REFERENCES






















































































AND ETHNICITY

Player A

Israelimale

Ashkenazicmale

Easternmale

Israelifemale

Ashkenazicfemale

Easternfemale Total

PLAYER

B

Ashkenazicmale

Easternmale

Ashkenazicfemale

Easternfemale

22

24

15

22

12

14

16

13

10

9

10

13

17

14

19

13

11

9

8

10

5

7

9

6

77

77

77

77

Total 83 55 42 63 38 27 308






















ETHNICITY

Player A

Israelimale

Ashkenazicmale

Easternmale

Israelifemale

Ashkenazicfemale

Easternfemale Total

PLAYER

B

Ashkenazicmale

Easternmale

Ashkenazicfemale

Easternfemale

23

22

15

19

10

14

10

12

14

19

13

9

26

18

13

14

17

16

10

8

10

11

9

8

100

100

70

70

Total 79 46 55 71 51 38 340






















the Trust Game





















BASED ON RANKS



Comparison of Z Plt





REFERENCES

















































































AND ETHNICITY

Player A

Israelimale

Ashkenazicmale

Easternmale

Israelifemale

Ashkenazicfemale

Easternfemale Total

PLAYER

B

Ashkenazicmale

Easternmale

Ashkenazicfemale

Easternfemale

22

24

15

22

12

14

16

13

10

9

10

13

17

14

19

13

11

9

8

10

5

7

9

6

77

77

77

77

Total 83 55 42 63 38 27 308






















ETHNICITY

Player A

Israelimale

Ashkenazicmale

Easternmale

Israelifemale

Ashkenazicfemale

Easternfemale Total

PLAYER

B

Ashkenazicmale

Easternmale

Ashkenazicfemale

Easternfemale

23

22

15

19

10

14

10

12

14

19

13

9

26

18

13

14

17

16

10

8

10

11

9

8

100

100

70

70

Total 79 46 55 71 51 38 340






















the Trust Game





















BASED ON RANKS



Comparison of Z Plt





REFERENCES






































































AND ETHNICITY

Player A

Israelimale

Ashkenazicmale

Easternmale

Israelifemale

Ashkenazicfemale

Easternfemale Total

PLAYER

B

Ashkenazicmale

Easternmale

Ashkenazicfemale

Easternfemale

22

24

15

22

12

14

16

13

10

9

10

13

17

14

19

13

11

9

8

10

5

7

9

6

77

77

77

77

Total 83 55 42 63 38 27 308






















ETHNICITY

Player A

Israelimale

Ashkenazicmale

Easternmale

Israelifemale

Ashkenazicfemale

Easternfemale Total

PLAYER

B

Ashkenazicmale

Easternmale

Ashkenazicfemale

Easternfemale

23

22

15

19

10

14

10

12

14

19

13

9

26

18

13

14

17

16

10

8

10

11

9

8

100

100

70

70

Total 79 46 55 71 51 38 340






















the Trust Game





















BASED ON RANKS



Comparison of Z Plt





REFERENCES
































































AND ETHNICITY

Player A

Israelimale

Ashkenazicmale

Easternmale

Israelifemale

Ashkenazicfemale

Easternfemale Total

PLAYER

B

Ashkenazicmale

Easternmale

Ashkenazicfemale

Easternfemale

22

24

15

22

12

14

16

13

10

9

10

13

17

14

19

13

11

9

8

10

5

7

9

6

77

77

77

77

Total 83 55 42 63 38 27 308






















ETHNICITY

Player A

Israelimale

Ashkenazicmale

Easternmale

Israelifemale

Ashkenazicfemale

Easternfemale Total

PLAYER

B

Ashkenazicmale

Easternmale

Ashkenazicfemale

Easternfemale

23

22

15

19

10

14

10

12

14

19

13

9

26

18

13

14

17

16

10

8

10

11

9

8

100

100

70

70

Total 79 46 55 71 51 38 340






















the Trust Game





















BASED ON RANKS



Comparison of Z Plt





REFERENCES





















































AND ETHNICITY

Player A

Israelimale

Ashkenazicmale

Easternmale

Israelifemale

Ashkenazicfemale

Easternfemale Total

PLAYER

B

Ashkenazicmale

Easternmale

Ashkenazicfemale

Easternfemale

22

24

15

22

12

14

16

13

10

9

10

13

17

14

19

13

11

9

8

10

5

7

9

6

77

77

77

77

Total 83 55 42 63 38 27 308






















ETHNICITY

Player A

Israelimale

Ashkenazicmale

Easternmale

Israelifemale

Ashkenazicfemale

Easternfemale Total

PLAYER

B

Ashkenazicmale

Easternmale

Ashkenazicfemale

Easternfemale

23

22

15

19

10

14

10

12

14

19

13

9

26

18

13

14

17

16

10

8

10

11

9

8

100

100

70

70

Total 79 46 55 71 51 38 340






















the Trust Game





















BASED ON RANKS



Comparison of Z Plt





REFERENCES






































































ETHNICITY

Player A

Israelimale

Ashkenazicmale

Easternmale

Israelifemale

Ashkenazicfemale

Easternfemale Total

PLAYER

B

Ashkenazicmale

Easternmale

Ashkenazicfemale

Easternfemale

23

22

15

19

10

14

10

12

14

19

13

9

26

18

13

14

17

16

10

8

10

11

9

8

100

100

70

70

Total 79 46 55 71 51 38 340






















the Trust Game





















BASED ON RANKS



Comparison of Z Plt





REFERENCES































































ETHNICITY

Player A

Israelimale

Ashkenazicmale

Easternmale

Israelifemale

Ashkenazicfemale

Easternfemale Total

PLAYER

B

Ashkenazicmale

Easternmale

Ashkenazicfemale

Easternfemale

23

22

15

19

10

14

10

12

14

19

13

9

26

18

13

14

17

16

10

8

10

11

9

8

100

100

70

70

Total 79 46 55 71 51 38 340






















the Trust Game





















BASED ON RANKS



Comparison of Z Plt





REFERENCES
























































ETHNICITY

Player A

Israelimale

Ashkenazicmale

Easternmale

Israelifemale

Ashkenazicfemale

Easternfemale Total

PLAYER

B

Ashkenazicmale

Easternmale

Ashkenazicfemale

Easternfemale

23

22

15

19

10

14

10

12

14

19

13

9

26

18

13

14

17

16

10

8

10

11

9

8

100

100

70

70

Total 79 46 55 71 51 38 340






















the Trust Game





















BASED ON RANKS



Comparison of Z Plt





REFERENCES







































































the Trust Game





















BASED ON RANKS



Comparison of Z Plt





REFERENCES































































the Trust Game





















BASED ON RANKS



Comparison of Z Plt





REFERENCES

























































the Trust Game





















BASED ON RANKS



Comparison of Z Plt





REFERENCES






































































BASED ON RANKS



Comparison of Z Plt





REFERENCES





























































BASED ON RANKS



Comparison of Z Plt





REFERENCES























































BASED ON RANKS



Comparison of Z Plt





REFERENCES



















































REFERENCES











































































Documents

DISCRIMINATIONINASEGMENTEDSOCIETY: …rady.ucsd.edu/faculty/directory/gneezy/pub/docs/... · 2012. 7. 5. · nationidenti”edinourtrustgameexperimentwasonlyapparent withrespecttoandbymaleplayers