Reducing meat consumption is considered a crucial and urgent element in terms of reaching climate targets, public health and the protection of animal welfare. Especially in industrialised countries, it is considered the most important recommendation in the field of sustainable food consumption. The transition is difficult to realise, as meat consumption is a complex area involving numerous institutions and stakeholders with conflicting interests and consumers with very different opinions and behaviour. Meat consumption is the result of diverse individual factors, while being rooted in culture and social norms. It is thus considered challenging for policymakers to set this transition in motion, as intervening in the system comes with uncertainties, a lack of understanding, and possibly resistance on both the stakeholder and consumer side. Simulation models can be supportive in such a case, as they can support an increased understanding of the system, while being able to deal with the transition and all its complexity. Specifically agent-based models can be a suitable tool to support policymakers in making sense of this transition, as this type of modelling can deal with heterogeneity of consumers and is able to provide insight on how various policy interventions affect the individual and overall system behaviour.
The process of developing a simulation model for policy support using a technique that is not widely understood and accepted yet, can be challenging. Participatory modelling is a method in which stakeholders are involved throughout the modelling process to promote social learning and achieve model improvement. In this thesis, a participatory modelling process was designed to identify the potential use of agent-based modelling in reducing meat consumption in the Netherlands. The study focused on including elements of behaviour in the ex-ante evaluation of policy with the use of an agent-based model.
An agent-based model representing meat consumption behaviour of Dutch young adults was evaluated and improved together with a group of participants. Knowledge elicitation with the stakeholders working in the field of policymaking, research, and academia occurred through interviews, workshops, and mind mapping. The key findings of these sessions were that the participants desire to gain understanding on the socio-cultural factors influencing meat consumption and how these can be targeted with interventions. To respond to this lack of knowledge, the agent-based model was adjusted to capture meat consumption behaviour according to the COM-B wheel. This is a theoretical framework in which behaviour is categorised into physical and psychological capability, reflective and automatic motivation, and physical and social opportunity. In the agent-based model, consumers select and consume meals from a supermarket, take-away, or restaurant. Dietary preferences are based on knowledge and skills, environmental and animal welfare concerns, income, desired meat consumption, and social norms. The consumers are able to adapt their dietary preferences through reflection processes. The individual profiles were empirically grounded with input data from a cross-sectional questionnaire.
The agent-based model was used to study the effectiveness of various policy pathways that were formulated in consultation with involved participants. The policy pathways included in this study were various meat price increases, an increase of the vegetarian representation in the food environment, a social marketing campaign targeted at social norms, and combinations of the three. Of these interventions, meat price increases showed to be most effective in reducing the overall meat consumption. Increasing the vegetarian representation in the choice was effective, but to a lesser extent. The social marketing campaign on social norms showed no direct effect on the amount of meat consumed. When interventions were combined, even higher reductions were observed.
This study sheds a light on how the field of policy and science can together work on a gained understanding of this complex transition, with the use of agent-based modelling.