Making More Sense of Agent-Based Simulation for Agricultural Policy Analysis

In the field of agricultural and resource economics there has recently been a growing interest in using agentbased models (ABM) for policy analysis. ABM possess the capability of simulating complex relationships between many interacting agents and their environment. In agricultural economics, ABM offer possibilities for addressing and explaining observable phenomena such as structural change. Many empirical based agent-based models are highly complex and include a multitude of modelled processes as well as a high degree of detail and parameterisation. This inevitably reduces their tractability, and makes it difficult to follow and understand their functioning and interpret results. Because of this, communicating results of complex agent-based models to policy-makers is a challenging task. For ABM to assist in decision-making, policy makers should develop an understanding of the complex processes and assumptions underlying the simulation models based on the provided given information (such as model documentations, model code). Yet, this is hardly a realistic option given policy makers' varying disciplinary backgrounds and time restrictions. Obviously, models cannot capture the full complexity of a target system and all relevant processes. Inevitably, we need to make guesses and assumptions about the true nature of the target system. However, we do not know what the response will look like if we for different combinations of input parameters, and how these interact with each other. This is particularly important, if we, for exmample, want to draw relevant policy conclusions based on an analysis of interactions between policy measures and determinants of structural change.