Anagrafe della ricerca

The project contributes to a novel chapter of mathematical control theory that studies models of opinion formation and decision making in social groups. The term “opinion”, in its broadest sense, stands for some numerical characteristic of a social agent (a person or an organization) that is altered by social influence, for instance, a cognitive orientation towards some object or event, an appraisal of another agent, a set of cultural or personal traits, the reputation. Dynamic models of opinion formation give simplified yet instructive description of many social processes and phenomena. We consider bounded confidence models on hypergraphs to model e-monetary networks where agents, characterized by homophily and biased assimilation, engage in many-body transactions. The paradigm of behavioral finance implies that opinion formation mechanisms, coming from social psychology, play a fundamental role in determining decisions of financial agents and interactions among them. However, applications of dynamic models in economics and finance are rather limited and primarily concerned with dynamics of markets and policies of targeted advertisement. A wide gap exists between dynamic models and complex behaviors of real financial agents. We aim to fill it by devising new models that are based on opinion formation and disclose key drivers of decision making in networks of financial agents. As a case study, we use a business-to-business (B2B) electronic currency network whose provider, Sardex S.p.A., has expressed its interest in the project and agreed to supply non-confidential data about the networks of transactions to be used for the model validation. The Sardex Ministero dell'Università e della Ricerca MUR - BANDO 2022 community includes thousands of members and provides its partners with a credit line used to purchase goods and services within the network. Recent works have revealed non-trivial topological features of the Sardex transaction network; however, the mechanism underlying the network evolution, which is driven by opinion formation, remains undisclosed. Opinions in this context characterize reputations of financial agents and influence the frequencies of transactions between them; conversely, the volume of transactions completed by an agent affects its reputation in a complex feedback loop between the opinions and the agent’s actions. A key feature of monetary networks is the presence of interactions in groups of three or more agents that cannot be reduced to sum of pairwise interactions. Sardex allows transactions with more than two agents, in which one agent can compensate for the purchase of an asset by closing a cycle of transactions with third parties. Therefore, we will make use of the paradigm of networks with higher-order interactions to capture the opinion formation mechanisms that reflect into the simple graph of transactions we observe. We expect that the analysis of our application will lead to a breakthrough in the theory of higher-order dynamical networks that is in its infancy.