The importance of being many: dynamics, interaction and aggregation in a multi-sector economy

This study develops a family of models to evaluate how agent heterogeneity and interactions shape macroeconomic dynamics, challenging the adequacy of representative agent frameworks. Building on dynamic multi-sector models populated by boundedly rational firms and households, we conduct both analytical and computational comparisons between aggregated and disaggregated representations across equilibrium and disequilibrium regimes. We identify precise conditions –individual and relational indistinguishability– under which representative constructs successfully replicate multi-agent dynamics, and we demonstrate their failure in constrained regimes where rationing-induced network shocks generate irreversible structural changes. The analysis reveals that aggregation errors escalate with heterogeneity, asymmetric interactions, and shock-driven reconfigurations of economic networks, critically undermining policy inferences. The proposed family of multi-agent models, grounded in minimal realistic principles, allows us to systematically quantify the errors derived by treating the response to exogenous shocks as a dynamic sequence of equilibria, rather than explicitly accounting for out-of-equilibrium dynamics. These insights bridge Keynesian coordination failures with modern complexity economics, offering methodological rigor to address Blanchard's critique on the relevance of interactions for macroeconomic modeling.