July 31, 2013 •Kelley MacEwen
In the most recent Econometrics Seminar Program session, Summer Analyst Larry Au provided an overview of Agent Based Modeling techniques and its various applications. Agent Based Modeling has a long history and can be traced back to the mid-20th Century to the works of John von Neumann. Further developments in Agent Based Modeling occurred through the fields of game theory and computational sociology. However, the use of Agent Based Modeling as a methodology did not take off until recent decades, with the advent of modern computing that enabled researchers to design and execute more robust models. Today, Agent Based Models have been deployed in fields as diverse as epidemiology to urban planning.
An Agent Based Model is made of several components. The first are the agent themselves, which are autonomous and heterogeneous in composition. Some programs enable agents to adapt and change their behavior and attributes throughout the simulation as they interact with one another. The second component is that of the environment, which imposes conditions on the agents and regulates the agents’ movement across its spatial boundaries. Finally, the models rely on either an asynchronous or synchronous schedule that dictates and times the co-movements of the agents and the environment. This microscopic view of the Agent Based Model allows complex phenomena to emerge from even the simplest rules.
The presentation looked into the applications of Agent Based Modeling in the study of spatial segregation, through the works of Thomas Schelling (1971), and network analyses of economics, such as those in research pioneered by James B. Glattfelder (2011) and Dirk Helbing (2013). Open source software available that supports Agent Based Modeling was reviewed as well. The session ended with a discussion on the broader applicability of Agent Based Modeling and its implication for the social sciences and public policy.
Below are two applications of Agent Based Modeling that Larry provided to the Seminar:
Two simulations of urban systems: (L) An implementation of the Schelling model created with NetLogo, (R) A complex model depicting housing conditions, transportation infrastructure, population growth, and economic dynamism developed by AnyLogic.
