all the more attractive to users (Kain, Vol. 2). With buses less delayed by traffic congestion, both drivers and buses could be used more productively.

Congestion pricing would reduce the demand for new highway capacity; this reduced demand would ease the capital requirements for expanding highways in response to growing population and travel demand. The U.S. Department of Transportation currently estimates that over the next 20 years the capital demands for the nation's roads and highways will range between $758 and $1,009 billion (in 1991 dollars) (DOT 1993). The low end of the range in costs represents the mere addition of sufficient capacity to maintain congestion at existing levels. The high end of the range represents the cost of easing congestion above existing conditions. The cost of expanding urban roadways in response to congestion will range between $232 and $365 billion. On an annual basis, the need for roadway expansion in urban areas totals $11.6 to $18.25 billion (of a total annual cost for urban and rural investment of $38 to $50.5 billion). Because of the phenomenon of latent demand, there is little prospect that these investments would actually reduce congestion. If congestion pricing were to be adopted nationwide, it would greatly reduce most of the new capital investment needs in urban areas (Winston and Bosworth 1992). It would result in time savings and reduced investment needs valued at roughly $5 to $11 billion (Small et al. 1989; Repetto et al. 1992).

Similar to any other marketlike mechanism for allocating demand, the fees imposed by congestion pricing would cost a larger proportion of the income of low-income motorists, and presumably these motorists would be among the most likely to shift travel time or mode to avoid the fee. Because of the substantial revenues that congestion pricing can raise, however, it is possible to provide compensation to those groups most disadvantaged by congestion pricing. For example, to offset adverse impacts on the poor, some of the revenues could be used to reduce the regressive taxes (property, sales, and gasoline) used to finance roads and transit.

The revenues earned from congestion pricing could be used to ensure that all income groups benefit; this is important, because without the benefits that these revenues provide, the average commuter would be somewhat worse off than before (Small 1983; Hau 1992). The time gains of