Assessing the effects of input uncertainties on the outputs of a freight demand model

Freight demand models typically use a series of sub-models that relate several inputs to model outputs, without considering the uncertainty of typical data sources used in the model development process. Hence, the uncertainty of freight demand model outputs is typically not characterized, making it difficult to understand the robustness of the model results, or how the rigour of the results might be improved with additional data. This paper proposes a formal five-step framework to analyze the effects and propagation of input uncertainty from datasets used during model development on the uncertainty of the outputs in a freight demand model driven by exogenous economic forecasts. The framework is applied to a Canadian commodity-based freight demand model, inspired by the Aggregate-Disaggregate-Aggregate (ADA) model, used to analyze the effects of the Comprehensive and Progressive Trans-Pacific Partnership (CPTPP) on Canada's trade infrastructure. In this application of the framework, uncertainty for input datasets used to develop three sub-models is introduced and a set of outputs is simulated through repeated simulation. Descriptive statistics and rank error measures are used to access the uncertainty of the outputs. The results suggest that the case study model performs with adequate robustness in terms of aggregated outputs and for larger trade partners, while some specific disaggregate outputs and scenarios with smaller trade partners are less robust.