Revalidating the Mechanisms of Reciprocity in Collaborative Networks

• In Event: Collaborative Governance for a Clean, Smart, and Resilient Society: Insights into Policy and Practice

Abstract

Motivation: Reciprocity is a cornerstone concept in social network analysis within the collaborative governance literature, frequently used to explain bonding behaviors in networks. However, the mechanisms driving reciprocity remain contested. Two competing theories dominate the literature: one attributes reciprocity to social capital norms, where trustworthiness fosters stronger ties, while the other links reciprocity to risk aversion, suggesting it is a strategic response to mitigate the risk of defection. Despite their distinct premises, these theories lead to similar predictions about network dynamics, making it difficult to empirically differentiate between them. This research employs a two-study approach to clarify these mechanisms. 

Study 1—Reciprocity in Two Actor Mode Behavioral Experiment: We designed a behavioral game experiment on 981 policymakers across more than 2,000 U.S. municipalities. This experiment uses a 2×2 factorial ultimatum game design to compare how social capital and risk aversion conditions affect inter-jurisdictional financial collaboration in climate resilience. Participants allocated a $1 million fund under conditions varying by social capital (low vs. high, based on the proportion of prior funds received) and risk aversion (high vs. low, based on the number of competing cities). 

Results indicate that both high social capital and low risk promote reciprocity, but social capital’s influence is 7.5 times stronger. Subgroup analysis shows that risk aversion only predicts reciprocity when social capital is low, while social capital effects persist regardless of risk levels. Exploratory analysis examines variations across political affiliations and government roles. 

Study 2—Reciprocity in Multiple Actor Mode Agent-Based Modeling: We then apply and extend this behavioral approach to simulated network activities that include more than two players. We developed an agent-based model simulating multi-agent interactions among public officials, where each agent is endowed with initial social capital and individual risk thresholds. Agents are embedded in diverse network topologies (e.g., random, small-world, scale-free) that mimic real-world inter-jurisdictional collaborations. 

In each simulation round, agents decide to reciprocate based on their updated social capital and a risk factor informed by local network density and competition; outcomes of these interactions dynamically update their social capital levels, thereby influencing future decisions and the formation of collaborative clusters.

Theoretical Contribution: This study is the first to combine behavioral and simulation experiments to examine the causal effects of social capital and risk aversion on reciprocity within collaborative networks, thereby advancing the theoretical understanding of reciprocity in interlocal collaborations.

Author

• Yixin Liu, Baruch College (City University of New York)
  Presenting Author