Agents First! Using Agent-based Simulation to Identify and Quantify Macro Structures

Abstract

The agent-based simulation approach models social phenomena as resulting from local interactions of agents one level below the phenomenon. Compared to system dynamics studies this often requires a lower level of abstraction what can make agent-based models easier to quantify. This paper analyzes the potential of the agent-based approach (i) to quantify a known aggregate causal structure and (ii) to identify the unknown aggregate structure of a problem. The two cases are investigated using very simple models of population dynamics.