APARTHEID SOUTH AFRICA

In an important study, one of the first of its kind in South Africa,1 Ashraf et al [2003] found black proposers to have significantly lower offers in a trust game, supporting previous work suggesting that members of previously disadvantaged groups in a society can trust less [Alesina et al, 2000]. The BDM [1995] version of the trust game was run, where proposers' offers were tripled, and once proposers made their offers, they were asked to record how much they expected their partner to return to them. Given the diverse locations of the participating schools, the offers were communicated between participants by cell phone by designated "phoners".

This protocol was followed in each group, and all telephone monitors used the same script when transmitting the offers to reinforce the credibility of the experiment in the eyes of the students. Although it is possible that this created some doubt in the students' minds about the credibility of the experiment, the post-experiment questionnaires where their doubts about the experiment were challenged do not indicate that this was a problem.7. In fact, black proposers expected, on average, to receive a greater share of triple the amount back than other students, and these expectations did not differ significantly by the race of the person with whom they were matched (Table 1).9 Furthermore, in matched students, the regression results in Table 2: Black bidders make significantly lower bids even after controlling for their expected returns.

9 Expectations of the mean proportion to be returned by black proposers are significantly higher than those for white proposers (15% significance using t-test) and colored proposers (significant at the 5% level using t-test; 11% using Mann-Whitney ).

Black Trustees Are Mistakenly Trusted Less

Thus, any expectation that Black filers would consistently return significantly lower amounts compared to other trustees is incorrect. Because black trustees make significantly higher return offers to proposers of color than do black proposers, it may be that black proposers in combination with black trustees rationally expected to receive lower return offers, and therefore made low offers to black partners in anticipation of this. Thus, black proposers' lower offers to black trustees may partly reflect this expectation, as suggested by the difference (albeit insignificant) in expected returns for black-black versus black-coloured pairs (Table 1).

However, even after expectations are taken into account, black filers still show a negative bias toward black trustees, so expectations alone cannot fully explain this result (Table 2).

White Proposers Prefer Not To Trust Blacks At All

Cragg's [1971] specification provides a means of testing whether the probability of a non-limiting outcome is determined independently of the level of the non-limiting outcome as a variant of the tobit model, using a likelihood ratio test. Using this specification, the likelihood ratio test confirms that for the regressions shown in Table 4, the probability of a threshold score is determined independently of the level of the non-threshold score.16 In addition, there are some interesting racial differences. For Black and Colored proposers, being paired with a Black partner has no bearing on the decision to make an offer or not.

On the contrary, it only affects (negatively and significantly) the size of the offer that is made. In stark contrast, white proposers are significantly less likely to bid with black partners, preferring to opt out of any interaction, while white proposers who decide to interact do not include black trustees treated significantly differently. - stricter than trustees from other racial groups.

Black And Coloured Trustees Are Less Trustworthy Towards Black Proposers

5 Racial Diversity and Trust

Racial Diversity In Schools And Peer Groups Is Positively Associated With Trust Towards Blacks

Interaction terms between racial diversity in schools and friendship groups are included to control for the possibility that despite attending racially mixed schools, students may still cluster with peers from the same racial group. The top panel of Table 8 presents the partial derivatives of these mean-estimated proposal offers, especially for proposers with black partners. 21 Contrary to expectations, it is not the case that the most racially diverse schools in this sample are bordered by Black and Colored neighborhoods.

Even after controlling for measures of racial diversity, black bidders still make significantly lower bids for black trustees than trustees from other racial groups. However, the extent of this mistrust is reduced for black proposers with same-race friends (Table 8, Row 1). In contrast, White and Black proposers with same-race friends have significantly less trust in Black administrators, while there is an associated reduction (and in the case of White proposers, a change) in this distrust of administrators black for those with mixed race friends.

However, proponents' behavior toward black administrators may vary depending on the extent of racial diversity in the proponent's school environment. Proponents' offers with Black partners increase the amount of racial diversity in their schools (Table 8, row 2), suggesting that exposure to a racially diverse environment is significantly and positively related to trusting behavior toward Black partners. This association is even greater if the proposer, especially black and white proposers, has mixed-race friends.

White and Colored introducers with same-race friends trust significantly less Black trustees than those with mixed-race friends, while the reverse is true for Black introducers (Table 8, Row 3). A key finding, however, is that racial diversity in schools and friendship groups interact in important ways. White and Black proposers in more racially diverse schools with same-race friends trust significantly less Black partners (Table 8, Row 4).

Racial Diversity In Schools And Peer Groups Is Positively Associated With Trustworthiness Towards Blacks

For proposers of color, the effect is in the opposite direction, a somewhat puzzling result that warrants further investigation in future work. White trustees in more racially mixed high schools with same-race friends are significantly less trusting of black proposers than white respondents in more racially diverse schools with mixed-race friends. This reflects the association between racial diversity in schools and friends and trusting behavior in white proposers.

6 Discussion

In doing so, these proposers are able to obtain important information that enables them to update their expectations and stereotypes regarding the reliability of black partners. An important and interesting extension of this work would be to observe interracial trust in a repeated trust game setting, which would allow one to measure the extent of the updating process for proposers of different racial groups paired with black partners. is, to observe. The systematic pattern of mistrust towards black partners can be characterized as "discrimination in contract" [Loury and although such discrimination may be morally objectionable, third parties such as the State must be used to prevent such discrimination from occurring in real world contexts. is definitely possible.

Interracial trust, especially toward blacks, is positively associated with increasing racial diversity in both the school environment and peer groups, both of which reduce "discrimination in contact." The racial composition of local networks, such as peer groups, may be particularly important, as individuals may continue to practice discrimination in contact with their close associates, despite attending racially diverse schools. However, to the extent that attending a more racially mixed high school increases the incidence of mixed-race friendships, as documented by Quillian and Campbell [2003], these two factors should work together over time to increase interracial trust.

7 Conclusion

Given the difficulties in conducting this type of experimental work on a large scale, small sample sizes are unavoidable, and there is an important need for further replication of this work, perhaps even using a repeated game setting to more carefully explore the idea of ​​racial stigma as opposed to racial stereotypes. Furthermore, if economists are serious about advancing our understanding of the mechanisms through which social capital, networks, and group relationships influence behavior, and thereby exchange, more detailed attention must be paid to collecting better and more comprehensive data on the composition and functioning of social networks. these types of institutions. The simple measures of racial and peer diversity used in this study, while compelling, are limited in illuminating the ways in which they actually influence behavior.

2002]: “Trust, Reciprocity, and Other Affective Preferences: Groups versus Individuals and Males versus Females,” in Advances in Experimental Business Research, ed. 1971]: "Some Statistical Models for Limited Dependent Variables with Application to the Demand for Durable Goods," Econometrica. Buchan [1999]: "Gender and Culture: International Experimental Evidence from Trust Games," American Economic Review [Papers and Proceedings.

Klitgaard [1998]: “Economic growth and social indicators: an exploratory analysis”, Economic Development and Cultural Change, XX [455-489]. Ver Boven[2002]: “Discrimination and nepotism: the efficiency of the anonymity rule”, discussion paper, Eitan Berglas School of Economics, Tel Aviv University. Moore [2001]: “Some doubts about measuring self-interest using dictator experiments: the costs of anonymity.” Journal of Economic Behavior and Organization.

Another Experimental Section on the Effects of Indian Caste on Performance," World Bank Policy Research Working Paper 3351, The World Bank and The Pennsylvania State University. Nagai[1996]: "Affirmative Action in Undergraduate Enrollment at the University of California, Berkeley A Preliminary Report" Center for Equal Opportunity, October; Sterling, VA. Campbell [2003]: "Beyond Black and White: The Present and Future of the Multiracial Friendship Divide," American Sociological Review, 68[August], 540–566.

Aronson[2002]: “Contending with Group Image: The Psychology of Stereotype and Social Identity Threat,” Advances in Experimental Social Psychology. In all of the above combinations, the petitioner's race is presented first and the trustee's race is presented second. The dependent variable, the amount offered in the trust game, is recorded to minimize scale effects.

The dependent variable is the fraction sent by the administrator (Amount returned/Triple amount received) and is not logged. Key: BlackProposer = Proposer is black; BlackTrustee= Trustee is black, RDSchool = Racial diversity in schools; RDFriends = Racial Diversity in Friendship Groups.

Table 1: Mean Offers And Returns In Trust Game By Race Of Participants

The Centre for Social Science Research

RECENT TITLES