The paper has produced a number of findings, of great importance, that call into question currentfreight demand modeling practices. The paper’s chief conclusion is that the accuracy of freight generation(FG) and freight trip generation (FTG) models depend on: the consistency between the model’s structureand actual FG/FTG patterns, the degree of internal heterogeneity of the economic/land use aggregationused to estimate the model, and the appropriateness of the spatial aggregation procedure used to obtainthe desired FG/FTG estimates.Regarding model structure, the paper establishes that there are strong practical and theoretical reasons totreat FG and FTG as separate concepts because the latter is the output of logistic decisions, while theformer is determined by the economics of production/consumption. As a result, the connection betweenbusiness size variables, e.g., employment, and FG is relatively strong because they are economic inputfactors; while the one with FTG is weaker as inventory and transportation costs come into play.Moreover, while FG is expected to grow with business size, FTG may not necessarily do the same. Thismeans that it is not generally correct to assume proportionality between FTG and business size, and thatusing constant FTG rates could be problematic. For instance, only 18% of the industry sectors in NewYork City exhibit constant FTG rates per employee.The paper also analyzes the role played by the economic/land use aggregation used to define the industrysegments included in a FG/FTG model, and the spatial aggregation for which the FG/FTG estimates aredesired. In terms of economic/land use aggregation the chief conclusion is that the finer its level of detailthe better, as independent variables would have better chance of explaining FG/FTG. In the case of spatialaggregation, the paper concludes that the correct aggregation procedure depends on the underlyingdisaggregate model. In essence, for a FG/FTG model to work well both economic/land use and spatialaggregations must be appropriate.
展开▼
