Three essays on demand for freight transportation: Optimization, spatial econometrics and parametric estimations.

摘要

The dissertation evaluates the demand for grain transportation using different methodologies such as a spatial Tobit demand estimation and the optimization models. The first essay, taking account of the spatial interactions, develops the Tobit demand system with different spatial weight matrices, which represents spatial effects on demand for transportation among neighboring elevators. I proposed the new methodology of systematically constructing an unequal weight matrix for the spatial Tobit demand model through the use of Geographic Information System (GIS). The main results indicate a significant negative impact of the spatial factor in the demand for rail transportation and heterogeneity among elevator companies.;The second essay examines the effects of Identity Preserved (IP) system on grain transportation in Washington by applying General Algebraic Modeling System (GAMS). I develop linear programming optimization models representing three scenarios, which are the current bulk grain transportation, the IP or containerized grain transportation and the IP system including extra material costs. The results indicate a significant change in wheat flows as a result of the IP system. The higher containerized transport rates and the prohibition of transshipment contribute to the rising transport costs. In addition, the sensitivity analysis of discounting the containerized rail rate identifies a spatial competition between the rail mode and the truck-barge mode.;The third study evaluates impacts of the Pacific Northwest (PNW)'s soft white wheat (SWW) marketing plan which aims to promote wheat to buyers by providing information of the different end-use qualities of the soft white wheat grown in different geographical areas. Under this model, I assume that elevators voluntarily move grain only within the production zone and there is no grain shipment across different production zones in an attempt to preserve identity of wheat grown in a specific zone. The cost-minimizing linear programming optimization model is applied to represent elevator managers' decision on grain transportation. The results indicate insignificant changes in wheat flows and a slightly increase in transport costs due to the imposition of the marketing plan. In contrast to the IP system, the zoning policy is significantly inexpensive as the result of the existing economies of scale.