Multioutput analysis of trucking operations using spatially dissaggregated flows

The operation of a trucking system at a national scale is analyzed by means of an operating multioutput cost function, which recognizes flows with different origin and destination as distinct components of the transportation product. The original, fully disaggregated, information on O-D specific flows is used to generate distance-weighted and pure-flow output vectors, whose dimension allows for the estimation of flexible cost functions. Different models are compared in terms of O-D specific marginal costs and second derivatives in common units, obtained through manipulations of both the estimated parameters and their variance-covariance matrix. Results indicate that marginal costs are not proportional to distance. Constant returns prevail in the operation of the system, simultaneously with economies of spatial scope and cost complementarity between flow pairs, with one exception which suggests some regional specialization. On methodological grounds, the bias on scale economies introduced by the single aggregated output measure (ton-km) becomes evident, and partial aggregation in terms of distance-weighted zonal flows seems more appropriate than pure flow aggregation when distances are heterogeneous. The multioutput formulation with O-D specific flows under a flexible form of the cost function, is recommended for meaningful policy analysis of transportation systems.