Using agent-based models to simulate political dynamics and pretest stakeholder engagement strategies
Agent-based models have become more sophisticated since Axelrod (1980) demonstrated that complete histories of villages could be simulated by assuming that the social actors, represented by modules of program code known as ‘agents’, have only a few basic motives and information processing rules governing their interactions with one another. This paper examines three applications relevant to dealing with a mine’s stakeholders. First, studies of the core motives and decision rules governing the dynamics that produce the ‘curse of natural resources’ are reviewed. The findings illustrate the benefits of viewing governments as ‘stationary bandits’. Second, studies examining information and attitude diffusion emphasize the importance of understanding which stakeholders are connected to which others. Third, the application of expected utility theory to stakeholder decision modelling illustrates that stakeholder political manoeuvering can be predicted when the outcome is one-dimensional (e.g., the level of social license granted). Examples from mines are presented to show the usefulness of the latter approach in pretesting stakeholder engagement initiatives before implementation. Directions for future development are discussed, including better use of existing secondary data sources and the improvement of base assumptions about what makes stakeholders interact as they do.