conditions. In metropolitan regions that are primarily concerned about controlling oxides of nitrogen, this implies that a pricing strategy would have to be very finely tuned to allow speeds to increase above stop-and-go conditions, but not so high as to allow free-flow conditions. Such fine tuning may not be practical with simple peak/off-peak pricing. Depending on how ongoing research reduces the uncertainty about the speed/emissions relationship and on the concern about oxides of nitrogen as opposed to other emissions, this could have implications for political feasibility by weakening the potential support of environmental groups.

Congestion pricing could reduce some ozone precursor emissions by simply reducing the actual number of trips taken. A large share of such emissions is caused by starting cold vehicles; hence a reduction in trips would reduce these emissions directly. About 50 percent of hydrocarbon emissions (reactive organics) result from cold starts and another 10 percent from evaporative emissions (the evaporation of emissions when the vehicle is not running but the engine is still hot).

In contrast to regionwide congestion pricing, estimates of emissions reductions from pricing individual facilities such as bridges or tunnels are highly dependent on the availability and capacity of alternate routes and on how much traffic would shift (Horowitz 1982). The availability of alternate routes may increase or decrease net emissions of carbon monoxide or oxides of nitrogen. To illustrate, consider two routes that serve the same origin and destination and that are reasonable substitutes for one another. Traffic on the two routes will tend to balance out as users shift back and forth to minimize trip times. If one of these routes becomes congested, some motorists will shift to the other. Alternatively, if one route were to be priced, demand for this route could be reduced during the peak and some traffic could shift during the peak to the alternate, unpriced route. Smoothing the traffic flow would reduce emissions of reactive organics or carbon monoxide for the priced route, but they could become worse on the alternate route if it became congested. These potential traffic shifts highlight the importance of designing congestion pricing pilot programs to avoid creating congestion at formerly uncongested points.

Regionwide congestion pricing would reduce energy consumption more directly than emissions. Fuel use would decline from both reduced trips and improved traffic flow. Harvey (Vol. 2) estimates a 6.5 percent reduc-