Using Agent-Based Models to Prototype Governance and Policy Designs

The ability of researchers and policy makers to envision how governance structures are configured and function can hinge on the extent to which design thinking is employed. “Design thinking” is the process of defining, ideating, prototyping and testing. Engineers and architects, among others, have employed this process to build bridges, homes, commercial building, machines, etc. The process of prototyping and testing creative new ideas and approaches has become engrained into their training and routine practices.

The conscious application of design thinking to the study and development of optimal governance structures is a relatively recent phenomena in public administration. Although one could argue that defining, ideating and even prototyping and testing have historically been conducted as “real world” policy experiments, the ability to apply such practices early in the development stages, as an architect does when approaching a new project, has not been widely considered to be possible. The evolution of new software, coupled with ever evolving theoretical design frameworks that provide complexity friendly, macro-micro scaling opportunities, is now allowing researchers and policy makers to employ design thinking tools before policies and programs get enacted. Scenario development and theory testing and tuning are possible through the use of simulations of models in concert with stakeholders. This paper provides an overview of one such tool, agent-based models (ABMs), and how it is being used to shed new insights into governance design, and in the process, bringing design thinking approaches into the policy and governance development process.

Agent-based models are computer models that allow autonomous agents to act and interact with each other and within wider environments. Agents may be construed as social actors (individuals, animals, groups of individuals, organizations), physical or natural objects (buildings, parcels of land, commodities, built and natural infrastructure), or socially constructed, but reified objects of organized activities: programs or projects or decision making processes. Agent behaviors are parameterized using fixed and variable values. Social agents may be endowed with certain gender, age, ethnicity, wealth, etc. Organizational agents may possess certain human resource capacity, financial resources, missions, objectives, sectorial characteristics, etc. Project agents may possess fixed resources or resource needs. Physical or natural agents, such as buildings and facilities or parcels of land may carry specific physical or natural characteristics. Agent parameters may also be variable, and subject to manipulation by the modeler to render specific scenarios.

Drawing on examples from the emerging literature on ABMs in policy and public administration, including ABMs developed to study water quality management and transportation project prioritization undertaken by the authors, this paper and presentation will provide those less familiar with ABMs an opportunity to learn about how they are structured, operate and used to consider optimal governance designs.