Forecasting urban highway travel for year 2005

As part of a study aimed at estimating suburban highway needs for year 2005, models were developed for forecasting daily vehicle miles of travel (DVMT) for urban areas and its distribution by highway functional class, urban location, and urban area size. A regression model combining both time series and cross-sectional data is used to establish the relationship between the per capita DVMT of 339 urban areas in the United States and a set of explanatory variables including real income, employment, number of persons per household, number of driver licenses per 1000 persons, a variable representing highway supply deficiency, and a time variable. The dynamic shift over time in share of travel between urban locations and highway functional classes as urban areas grow in size is represented by conditional logit models. This paper presents the major findings from the forecasting and distribution models for urban highway travel in year 2005. 30 refs., 3 figs., 9 tabs.