The use of utility functions in dynamic spatial simulation models for regional and urban planning has advanced considerably in recent years. However, the influence of different utility function formulations on simulation results remains unclear. To address this knowledge gap, this study implements three different utility functions in an agent-based regional accessibility model where individuals choose destinations to maximize the benefits gained from travel. Simulation results are then analyzed via an ANOVA test to assess whether they are sensitive to the utility function formulations. The analysis is complemented by observations of emergent system behavior based on the patterns of preferential destination choice to support the interpretation of these results. This paper details how the chosen utility function formulation influences simulation outcomes, and, as a consequence, conclusions drawn from dynamic spatial simulation models. The results of the linear additive weighted sum function and the Cobb-Douglas function exhibit similar behavior, unlike the CES-based model. In turn, the latter depicts stronger linear relationships with some of the response variables in the regression analysis. The presented findings contribute to the discussions on the implications of seemingly minor variations in key model design decisions of dynamic spatial models. The study emphasizes the need for transparent and well-justified model design choices to ensure the credibility and reliability of these models.
Victoria Kazieva (2026) studied this question.