Computational models of global and local interactions: Essays on binary choice decisions, network effects on self-employment and strategic voting dynamics

This PhD dissertation consists of three chapters that discuss the effects of global and local information on agents' choices. The first chapter studies conformity effects on individual choice when both local and global information are present. A standard discrete choice model that incorporates the social interaction effect to assign choice probabilities to agents is studied. The model is analyzed by using computational simulations. Agents are dispersed in a two dimensional toroidal lattice and they can gather information either from their von Neumann neighbors or from the whole community of agents. Agent heterogeneity is introduced through diversity in private incentives among agents. The main results of the simulations show that the effects of global and local information can produce conflicting informational data streams in agents, making it likely that unstable and volatile aggregate choice emerges. Likewise, the results indicate that the interaction between global and local information affects the shape of the distribution of the size of informational cascades. The model and its results are particularly useful to understand opinion dynamics. The second chapter analyzes social interactions effects on self-employment activities in Mexico City. A census tract based model that instruments conformity effects as the mechanism of contagion in the social network is studied. Two specifications of the model are analyzed: one is based only in local information and the other introduces a threshold that depends on local and global information. In general, the threshold model reproduces better the observed spatial clusters of self-employment at census tract level than a simple model of local interaction. This implies that threshold dynamics can be present when self-employment activities are subject to social interactions in economic environments where informality is strong. The third chapter proposes a revisited model of strategic voting with three parties where strategic voting occurs only when the voting share of the preferred party that a voter is willing to abandon is below a level of tolerance. In this model, voters make expectations naively with the last available electoral-poll. Results illustrate that strategic voters switch across basins of attraction under fair assumptions which implies that electoral dynamics can alternate between semi-Duvergerian equilibria.