Agglomeration dynamics in Regional Innovation Systems (RIS) are considered important determinants of economic direction and resilience. Yet, existing studies lacked tools to model how agglomeration mechanisms interact dynamically to shape innovation outcomes, specifically MAR and Jacobs externalities. Traditional methodologies, such as regression or case studies, cannot capture feedback loops and causality effects, leaving researchers without
actionable insights for innovation development. In this research, I aimed to materialise the mechanics of agglomeration patterns in Regional Innovation Systems to build a dynamic understanding of their effect on innovation performance anchored in the EU regional context. The final aim is to shape research-informed decision-making with the findings of this dynamic study.
To bridge this gap, I created an SD model to simulate agglomeration dynamics in RIS. Modelling agglomeration patterns requires different levels of granularity from traditionally aggregated innovation system modelling. To satisfy this requirement, I drew concepts from a thorough theoretical literature review. I analysed collected theories by making use of Williamson’s Institutional model, informing the exploratory aspect of the SD modelling. To verify the model, boundary adequacy, structure verification, dimensional consistency, parameter verification and extreme value tests were accomplished. To validate model behaviour, a replicative data validation, an expert validation and a parameters sweep were realised. The model was parametrised with Ile-de-France and Bretagne data to simulate
the interactions between different sectors of Jacobs and MAR RIS to inform system behaviour. Finally, a Sobol sensitivity analysis informed influential system variables with a multivariate parameter variation, subsequently reinforced by a scenario test. These tests have been analysed for both MAR and Jacobs regions.
The results of the modelling informed that the MAR model was more robust to uncertainties. It also showed a faster recovery from shocks than Jacobs regions. Additionally, government funding did not play a significant role in regional knowledge development. It therefore appeared more important that regional sectors developed links with nearby industries, whether in MAR or Jacobs regions, to ensure higher flexibility in periods of uncertainty. Decision-making can therefore examine ways to increase industry proximity in Jacobs regions and firm collaboration in MAR regions.
Finally, it highlighted the importance of timing in innovation generation, often overlooked in agglomeration studies.
Beyond the findings, the contribution of this thesis also reflected itself in the modelling, a further conceptual offering to understand the complexity of innovation development. While the data and assumptions are interesting considerations brought by this thesis, the strongpoint stood in the formalisation and pragmatism of field theories through causal relationships in a model. Theoretically, it advanced RIS scholarship by reframing agglomeration as a systemic equilibrium rather than a linear outcome, while methodologically, it demonstrated how conceptual SD can democratise policy design.
The model offered can be further refined, detailed and parametrised in further research to lead to more robust outcomes and more precise data-informed decisions. One key improvement suggested is the inclusion of Agent Based Modelling to form a hybrid-SD-ABM model. Representing variables at the levels of single agents can support higher flexibility and dynamism, which can be advantageous to represent regions with a start-up landscape more
accurately.