Pareto Improvements from Lexus Lanes

This paper shows that a judiciously designed toll applied to a portion of the lanes of a highway can be a Pareto improvement even before the revenue is spent. I achieve this new result by extending a standard dynamic congestion model to reflect an important additional traffic externality which transportation engineers have recently identified: additional traffic does not simply increase travel times, but can also introduce additional frictions that reduce throughput. By using a time varying toll to smooth the rate that people depart for work it is possible to avoid these frictions, increasing speed and throughput. Because adding a toll increases throughput, if road users are homogeneous then adding tolls to all of the lanes is a Pareto improvement. However, adding tolls changes the currency used to pay for use of the highway during rush hour from time to money. This change hurts the inflexible poor. We can avoid hurting the inflexible poor by only adding tolls to a portion of the lanes. Doing so preserves their ability to pay with time instead of money. I show that as long as there are some rich drivers using the highway at the peak of rush hour then adding tolls to a portion of the lanes is a Pareto improvement. To confirm the real world relevance of this theoretical possibility I use survey and travel time data from California State Route 91 to estimate the effects of adding optimal time varying tolls. I find that adding tolls to a four of the lanes is a Pareto improvement, and that social welfare gains of doing so are over a thousand dollars per road user per year.