This paper is a case study of carpooling behavior on the 91 Express Lanes. The 91 Express Lanes are the nation's first implementation of High Occupancy/Toll (HOT) lanes where carpools with three or more passengers could use the lanes for free (at the time the data for this study was collected) and others pay a toll that varies by time of day to use the premium Express Lane. One concern over such a policy is that people won't carpool if they can just pay for the travel time savings that they would normally obtain by carpooling and using a High Occupancy Vehicle (HOV) lane. Our survey data show that the rate of carpooling did not change much between the opening of the Express Lanes and now, there is a lot of changing between modes (increases and decreases in the number of passengers), there are a large number of people that carpool a few times a week, and that HOV-2s use both the regular lanes and the Express Lanes. We further investigate whether HOT lanes encourage carpooling by modeling carpool formation with discrete choice models. The results show that mode choice behavior in the corridor is similar to carpooling behavior in other locations and carpooling in the corridor is not discouraged.

Transportation professionals have always been interested in how travelers respond to different transportation options. A new application of congestion pricing offers the opportunity to extend such research to situations where travelers face a priced alternative. Travelers along State Route 91 (SR 91) in Southern California can now pay a time-varying fee in order to travel on a set of essentially congestion-free "Express Lanes" located in the median of a very congested preexisting freeway. For this study, we conduct a mail survey of such travelers to learn how they decide to use the free lanes or the toll lanes. We use the data to estimate route choice models and models that incorporate various types of real-time information about accidents, traffic conditions, and price levels into the route choice decision. This study provides new information about the acceptance of congestion pricing, the use of real-time information in making dynamic travel decisions, and individual travelers' interests in forming carpools.

We develop a traffic congestion index using data for California highways from 1976 through 1994. The technique yields a congestion measure which has several advantages. The index developed here can be applied to counties, urbanized areas, highway segments, or other portions of geographic areas or highway networks. The index allows cross-sectional and time series comparisons which have only rarely been possible. Most importantly, the congestion index developed here is based on data which are readily available. We compare our index to others based on Highway Performance Monitoring System (HPMS) data, and illustrate similarities and differences. We also discuss important issues for future research and data collection efforts which can contribute to more refined congestion measurement.