In this paper, we design an economic experiment aiming at capturing the main traits of upward social mobility and measure its impact on behaviors in a trust game (Berg et al., 1995).
Upward mobility is a form of social mobility according to which an individual from a low status group is promoted to a high status group. In some developed societies, this upward mobility is still relatively rare and inherited group membership strongly predicts one's nal position in the social hierarchy. An example of upward mobility is a child from a blue-collars family who ends up surgeon. While this situation is possible, it remains rare.
Upward social mobility has long been considered a cornerstone of modern democracies because preserving some opportunities for every members of the society is pivotal in preserving social stability (de Tocqueville, 1835;Acemoglu et al., 2016), and because it's a fundamental component of a widespread notion of justice (Sen,1980). In this paper, we focus on the eects of upward mobility on a specic expression of social preferences : interpersonal trust. Trust is a very important dimension of social capital, because in most economic interactions, agents have to trust each others (Arrow, 1972).
Interestingly, while trust and social mobility seem both desirable, there is some evidence outside economics that upward mobility disrupts social preferences. For instance, some papers in social psychology suggest based on survey data and qualitative analysis that mobile individuals are often disapproved by left-behinds and not considered as equal by the members of the achieved group (Derks et al., 2015; Kulich et al., 2015). Then, the aim of our paper is to assess whether there is a tension between upward social mobility .and trust.

Experimental design
We design an economic experiment in which we generate upward mobility, and measure its impact on interpersonal trust. We build on social psychology conceptualization of upward mobility, which can be thought of as the discrepancy between one's original and one's achieved group membership. We use a natural group affiliation to generate original group affiliation and we induce the achieved group membership within our experiment. There are two original groups and two achieved groups. Every participant is affiliated with one original and with one achieved group. Being affiliated to one of the two achieved group provides a greater status than being affiliated to the other. Consistently with the denition of upward mobility, the original group affiliation is a good predictor of the achieved group affiliation. In practice, the experiment is divided in two parts :
In the first part, we generate social mobility. The original group affiliation is done by recruiting participants from the local business school (EM Lyon) and engineer school (EC Lyon). While being similar in prestige and in educational requirements, those two schools differ in their curriculum : selection and education in the engineer school is more math-oriented. We generate achieved group aliation by assigning participants to one of two roles : experts or agents. This labeling generates a difference in status, reinforced by a task in which experts have to evaluate the quality of agents' work. We assign participants to one of the two achieved groups depending on their performances in a math quiz. Participants from the engineer school are expected to, and do perform better, so that they represent the majority of experts. By design, we insure that some participants from the business school who performed well in the math quiz are affiliated to the experts group. In our experiment, participants can have one out of four possible identities, defined as two-dimensional group affiliation :

1 - Businness school & Agent (left behinds)

2- Business school & Expert (Upward mobile )

3- Engineer school & Agent (Downard mobile)

4- Engineer school & Expert (Expected expert)
 
In the second part, every participant takes part in four consecutive trust games as truster and as trustee. Each trust game corresponds to a different match. For each match, both players know both the original and the achieved group affiliation of his counterpart. We use the strategy method for the decision of the trustee and give no feedback between matches in order to minimize learning. For each trust decision, we elicitate the truster's beliefs about how much the trustee will reciprocate in an incentive-compatible
manner. Our main variable of interest are trust and reciprocity in matches that include an upward mobile individual. 

Preview of the results

Data show that upward social mobility does have an impact on trust and reciprocity. First, the upwardly mobile individuals trust left-behinds less than other left-behinds do. Surprisingly, we also show that upwardly mobile individuals are more reciprocal toward left-behinds than other left-behinds. This means that the decrease in trust is not due to a decrease in unconditional kindness. Moreover, our data suggest that the channel of beliefs does not explain this pattern. Our interpretation is that mobile individuals
are intrinsically more reluctant to let their fate in the hands of left-behinds. They anticipate a greater disutility in case left-behinds don't reciprocate their trust. In the wording of Bohnet et al. (2008) or Aimone and Houser (2012), mobile individuals are more averse to betrayal than non mobile individuals.

Second, while left-behinds show in-group favoritism and clearly discriminate against participants from the engineer school in terms of trustworthiness, we cannot reject the null hypothesis that mobile individuals are equally reciprocal toward engineers and left behinds. Then, upward mobility seems to decrease one's in-group favoritism.
On the other hand, our data shows that the left-behinds do not behave toward upwardly mobile individuals differently than they do toward other left-behinds. Nevertheless, engineers tend to trust upwardly mobile individuals less than they trust left behinds. The channel of beliefs doesn't explain this pattern, so that we argue that unconditional trust is affected : for a given level of expected reciprocity, engineers want to transfer less to upwardly mobile individuals than to left behinds. This is surprising because one could expect that expert engineers would feel closer to mobile individuals than to left behinds and then trust them more. This pattern can be understood in the
framework of social identity theory (Akerlof and Kranton, 2000) : it predicts that people who depict traits that are untypical from their group expose themselves to sanctions.

References

Acemoglu, D., G. Egorov, and K. Sonin (2016, April). Social mobility and stability of democracy: Re-evaluating de tocqueville. Working Paper 22174, National Bureau of Economic Research.
Aimone, J. A. and D. Houser (2012). What you don't know won't hurt you: a laboratory analysis of betrayal aversion. Experimental Economics 15 (4), 571-588.
Akerlof, G. a. and R. Kranton (2000). Economics and identity. The Quartely Journal Of Economics 115 (3), 715-753.
Arrow, K. J. (1972). Gifts and exchanges. Philosophy and Public Affairs 1 (4), 343-362.
Ashraf, N., I. Bohnet, and N. Piankov (2006). Decomposing trust and trustworthiness. Experimental Economics 9 (3), 193-208.
Berg, J., J. Dickhaut, and K. McCabe (1995). Trust, reciprocity, and social history. Games and Economic Behavior 10 (1), 122 -142.
Bohnet, I., F. Greig, B. Herrmann, and R. Zeckhauser (2008, March). Betrayal aversion: Evidence from brazil, china, oman, switzerland, turkey, and the united states. American Economic Review 98 (1), 294-310.
de Tocqueville, A. (1835). De la démocratie en Amérique (Democracy in America) Union générale d'éditions.
Derks, B., C. van Laar, N. Ellemers, and G. Raghoe (2015). Extending the queen bee effect: How hindustani workers cope with disadvantage by distancing the self from the group. Journal of Social Issues 71 (3), 476-496.
Kulich, C., F. Lorenzi-Cioldi, and V. Iacoviello (2015). Moving across status lines: Low concern for the ingroup and group identication. Journal of Social Issues 71 (3), 453-475.
Sen, A. (1980). Equality of What? Cambridge: Cambridge University Press. Reprinted in John Rawls et al., Liberty, Equality and Law (Cambridge: Cambridge University press, 1987).