Exploring policy change through agent-based simulation

Policymaking is a complex process that has been studied using policy process theories almost exclusively. These theories have been built using a large number of qualitative cases. Such methods are useful for theory building but remain limited for theory exploration and policy advice. On a different front, socio-technical system (STS) simulations are often used to test the impact and effectiveness of policies. This is done by using policy scenarios. This manner of dealing with deep uncertainty disregards the dynamic aspect of the policy process and of the way STS interact with policies. The objective of this dissertation is to present an approach that can be used to systematically model and simulate the policy process. This dissertation demonstrates how such a model can be used in these two widely different applications to explore the policy process theories and to better account for deep uncertainty in STS simulations. In the first part of the dissertation, I present a common language. It considers four building blocks using concepts from prominent theories as requirements to any model of the policy process: time, the policy arena, actors and the environment, and the actor interactions. Additionally, in this part, I argue why agent-based modelling is best suited to simulate the policy process. Finally, I present the hybrid modelling approach that is used to incorporate the policy process model into already existing STS models. The second part focuses on the simplest implementation model. This model is the first use of the common language and is entitled simplest implementation as the common language is used to build the simplest model possible capable of emulating the policy process. It is created to demonstrate the potential of the common language and is tested with three different STS. A predation model is used to present the possibilities stemming from this simplest implementation model. It is then added to a model of the Swiss electricity market to demonstrate what benefits the endogenisation of the policy process can have on the study of a complex STS. As a third example, the policy process model is added to a system dynamics model of flood safety in The Netherlands to demonstrate its versatility. The third part of this dissertation consists of the advocacy coalition framework (ACF) implementation model. In this part, I present a model, also developed using the common language, that is aimed at emulating the ACF as closely as possible. The ACF being a complex framework, this implementation is done in steps of increasing complexity. The first step introduces policy learning to the model, the second introduces coalitions, the third and fourth steps introduce aspects of partial knowledge and imperfect information. This ACF implementation model is also demonstrated using first a predation model and second a model of the Swiss electricity market. In conclusion, this dissertation shows that it is possible to model the policy process. However it was shown that the policy process theories could benefit from additional and more focused studies so that behaviours and mechanisms can be better understood. This work can be the first step to such studies. It was also shown that introducing the policy process in a STS simulation can be useful to better understand said system. This approach allows for an easy integration to already existing models. Finally, this work can be extended by looking at other theories or different applications.