Abstract
This paper describes a development of national freight demand models for 27 industry
sectors covered by the 2002 Commodity Flow Survey. It postulates that the national
freight demands are consistent with U.S. business patterns. Furthermore, the study
hypothesizes that the flow of goods, which make up the national production processes of
industries, is coherent with the information described in the 2002 Annual Input-Output
Accounts developed by the Bureau of Economic Analysis. The model estimation
framework hinges largely on the assumption that a relatively simple relationship exists
between freight production/consumption and business patterns for each industry defined
by the three-digit North American Industry Classification System industry codes
(NAICS).
The national freight demand model for each selected industry sector consists of two
models; a freight generation model and a freight attraction model. Thus, a total of 54
simple regression models were estimated under this study. Preliminary results indicated
promising freight generation and freight attraction models. Among all models, only four
of them had a R2 value lower than 0.70.
With additional modeling efforts, these freight demand models could be enhanced to
allow transportation analysts to assess regional economic impacts associated with
temporary lost of transportation services on U.S. transportation network infrastructures.
Using such freight demand models and available U.S. business forecasts, future national
freight demands could be forecasted within certain degrees of accuracy. These freight
demand models could also enable transportation analysts to further disaggregate the CFS
state-level origin-destination tables to county or zip code level.