Abstract
A new framework for inter-regional commodity flow forecasting is presented to improve estimates of freight demand for inter-regional and statewide transportation models. The Structural Equations for Multi-Commodity OD Distribution (SEMCOD) model is based on simultaneous direct demand equations with structural relationships between dependent and independent variables of the model. SEMCOD is a flexible model that integrates the generation and distribution steps in conventional four-step demand models. This integration provides consistent estimates for elasticity analysis of effective factors for freight flows at the OD level and for productions and attractions at the zone level. Also, the model is sensitive to policies that increase or decrease generalized transportation cost, not only for flow distribution but also by measuring the change in marginal production and attraction of each zone. Unlike gravity-type models, this framework provides the opportunity to identify homogenous clusters of ODs and to more accurately estimate parameters for each cluster. The proposed model is estimated using the Freight Analysis Framework (FAF3) and other publicly available data sources for 15 commodity groups. Elasticity of different factors on production, attraction and flow of different commodity groups with respect to industry specific employment, population, industrial GDP, variables related to consumption and production of energy and land use variables, are studied. Considering cross relationships between supply chains of different commodity groups in the model significantly improved the fitness of the model. The fitness measures confirm satisfactory performance of the model.