Advancing freight demand prediction: Unified destination and mode choice models in extra-urban transportation

Abstract

Freight transportation modeling represents the economic demand for moving goods across spatial networks. Traditional models often treat transportation mode and destination choice as separate processes. This paper introduces an innovative framework that synthesizes a joint choice structure with a novel, supply-chain-based methodology, an approach underutilized in freight modeling. The core methodological innovation is the use of national input-output accounts to identify specific consumer industries, allowing for a more behaviorally-grounded analysis of destination attractiveness for different commodity groups. Using U.S. Commodity Flow Survey (CFS) data, two multinomial logit models were developed: one for destination choice and another for joint destination-mode choice, across three goods categories. This study also provides a direct empirical comparison in the freight context, with results revealing that the integrated model improves prediction accuracy by approximately 7 % compared to destination-only models. The findings confirm the value of this synthesized approach, which provides a more comprehensive and behaviorally sound tool for freight demand estimation.