Agent-Based Modeling as a Legal Theory Tool

Sebastian Benthall, Katherine J. Strandburg

Agent-based modeling (ABM) is a versatile social scientific research tool that adapts insights from sociology and physics to study complex social systems. Currently, ABM is nearly absent from legal literature that evaluates and proposes laws and regulations to achieve various social goals. Rather, quantitative legal scholarship is currently most characterized by the Law and Economics (L&E) approach, which relies on a more limited modeling framework. The time is ripe for more use of ABM in this scholarship. Recent developments in legal theory have highlighted the complexity of society and law’s structural and systemic effects on it. ABM’s wide adoption as a method in the social sciences, including recently in economics, demonstrates its ability to address precisely these regulatory design issues.

The time is ripe for legal scholars to use agent-based modeling (ABM) to produce actionable theoretical insights. One major strand of legal scholarship attempts to design or evaluate potential regulatory approaches based on their anticipated effectiveness at achieving societal goals [1, 2, 3]. This is an inherently normative project: both the overall consequentialist perspective and particular choices of appropriate goals are contestable. This sort of legal scholarly project also has social scientific underpinnings, however, because its success depends on the quality of its predictions about how society will respond to legal changes. Indeed, legal scholarship of this sort is often explicitly interdisciplinary, relying on theoretical concepts, models and methods from various fields of social science fields to inform those predictions. Microeconomics has been particularly influential, spawning a sub-field known as “law and economics,” (“L&E”) or sometimes “economic analysis of law” [4–7]. (As one rough measure of this impact, the LEXIS database of law journal articles contains more than 35,000 articles mentioning “law and economics” or “economic analysis of law,” just over 3,000 articles mentioning “law and sociology” or “sociology of law” and just about 1,500 mentioning each of “law and political science” and “law and psychology.”)

ABM is a computer simulation approach that has been increasingly deployed in social science to study the societal implications of various specifications of agents (who can be modeled as individuals, firms or other entities), their incentives and decision-making strategies, the interactions between those agents, and the social frameworks in which they interact. The computer simulation approach allows agent-based models to incorporate heterogeneity, nonlinearity and feedback effects in ways that are not possible with more traditional analytical solutions and approximation techniques [8, 9]. Using ABM’s bottom-up “generative social science” approach, “fundamental social structures and group behaviors emerge from the interaction of individuals operating in artificial environments under rules that place only bounded demands on each agent’s information and computational capacity” [10].

ABM’s strengths would appear to make it an attractive approach for exploring the potential societal implications of proposed changes to laws and regulations, particularly in light of the difficulty of employing experimental methods to probe these issues. Moreover, laws and regulations intersect with many aspects of social life that have been studied with ABMs. Yet the legal literature seems surprisingly oblivious to ABM’s potential to inform the evaluation of proposals for legal and regulatory change. Because ABM is an increasingly important social scientific tool, its lackluster uptake by legal scholars who aim to predict the effects of regulatory proposals is disappointing. Nonetheless, we believe that several developments make a more robust incorporation of ABM into legal scholarship possible now. A first set relates to legal scholarly demand for less individualistic and more systemic, structural, and political approaches to regulatory design, while a second set relates to the legal academy’s capacity for and openness to computational modeling. Both sets are usefully understood in relation to “law and economics,” which has been one of the most (arguably the most) influential–and controversial – strands of legal scholarship since the seminal work by Posner and others in the mid-1980s [6, 7, 11].

Download the PDF

Engelberg Center Authors

Katherine J. Strandburg