Please cite this article in press as: Bravo, G., et al., Trust and partner selection in social networks: An experimentally grounded model. Soc. Netw. (2012), doi:10.1016/j.socnet.2012.03.001 ARTICLE IN PRESS G Model SON-715; No. of Pages 12 Social Networks xxx (2012) xxx–xxx Contents lists available at SciVerse ScienceDirect Social Networks journa l h o me page: www.elsevier.com/locate/socnet Trust and partner selection in social networks: An experimentally grounded model Giangiacomo Bravo a,∗ , Flaminio Squazzoni b , Riccardo Boero c a Università degli Studi di Torino, Collegio Carlo Alberto, and GECS-Research Group in Experimental and Computational Sociology, Italy b Dipartimento di Studi Sociali, Università degli Studi di Brescia, and GECS-Research Group in Experimental and Computational Sociology, Italy c Dipartimento di Science Economiche e Finanziarie “G. Prato”, Università degli Studi di Torino, and GECS-Research Group in Experimental and Computational Sociology, Italy a r t i c l e i n f o Keywords: Partner selection Trust Cooperation Laboratory experiment Agent-based model Social networks a b s t r a c t This article investigates the importance of the endogenous selection of partners for trust and cooperation in market exchange situations, where there is information asymmetry between investors and trustees. We created an experimental-data driven agent-based model where the endogenous link between interaction outcome and social structure formation was examined starting from heterogeneous agent behaviour. By testing various social structure conﬁgurations, we showed that dynamic networks lead to more coopera- tion when agents can create more links and reduce exploitation opportunities by free riders. Furthermore, we found that the endogenous network formation was more important for cooperation than the type of network. Our results cast serious doubt about the static view of network structures on cooperation and can provide new insights into market efﬁciency. © 2012 Elsevier B.V. All rights reserved. 1. Introduction Recent experimental studies have acknowledged the relevance of trust for cooperation in human societies (e.g., Barrera and Buskens, 2009; Berg et al., 1995; Boero et al., 2009a,b; Camerer, 2003; Keser, 2003). From a sociological perspective, the problem with these studies is that they considered only highly unrealis- tic interaction structures, e.g., random coupled subjects, and did not investigate the endogenous forces that might inﬂuence struc- ture formation. On the other hand, certain formal models with more concrete sociological content have examined how structural embeddedness affected trust and cooperation, but without refer- ring to empirically veriﬁed assumptions on agent behaviour (e.g., Cohen et al., 2001; Pujol et al., 2005). This article aims to study the link between interaction outcomes and social structure formation by combining experimental data on individual behaviour and social network simulation. Unlike certain inﬂuential studies that emphasized the relevance of the interaction continuity to understand cooperation (e.g., Axelrod et al., 2002; Cohen et al., 2001), we investigated whether the robustness of cooperation depended on the agent’s capability of selecting good partners, whether this could be inﬂuenced by the particular network structure where agents were embedded, and what was the effect of endogenous forces of network formation. ∗ Corresponding author at: Dipartimento di Scienze Sociali, Via S. Ottavio 50, 10124 Torino, Italy. Tel.: +39 011 670 26 35. E-mail address: giangiacomo.bravo@unito.it (G. Bravo). Our aim was to tackle certain important and unsolved issues in the experimental and simulation literature. First, there is no consensus about the role that network structures play on coop- eration. On the one hand, some authors argue that conditional cooperators could reinforce pro-social behaviour when they are linked with each other and cooperation could even increase when network structures convey social contagion (e.g., Fowler and Cristakis, 2010). On the other hand, others have argued that network structures could cause reciprocal defection and that the tightness required by social networks to favour contagion cascades was highly unrealistic (Suri and Watts, 2011). Further- more, most experimental and simulation studies on networks and cooperation have not considered under what circumstances and forces individual behaviour and social network conﬁgura- tions might inﬂuence each other endogenously (Takács et al., 2008). To ﬁll this gap, we have extended the scope of a standard exper- imental economics game, i.e., the repeated investment game, to situations that were impossible to test in the lab, by introducing complex network structures endogenously. Our results show, ﬁrst, that exogenously ﬁxed network structures do not lead to more cooperation than the experiment where subjects played in ran- dom dyadic networks. This conﬁrms Suri and Watts (2011), whose results indicated that, by varying the network structure, no sig- niﬁcant differences were found in contribution levels in a public good game. More importantly in our case, a substantial increase in cooperation occurred only when partner selection and dynamic structure formation were endogenously introduced. This allowed cooperators to break their links with free-riders, beneﬁt from a 0378-8733/$ – see front matter © 2012 Elsevier B.V. All rights reserved. doi:10.1016/j.socnet.2012.03.001