Using very detailed data from the Caltrans Performance Measurement System (PeMS) database, Varaiya has documented the massive deterioration in freeway performance that occurs during rush-hour congestion. Basically, beyond a critical level of traffic flow, adding more vehicles sharply reduces flow and speeds, leading to stop-and-go congestion. In one much-cited example on the I-10 westbound, a flow of 2,100 vehicles/lane/hour at 58 mph collapses to 1,300/lane/hr. at 13 mph—a drop in highway efficiency to 15% of the free-flow level. The data show that low levels of efficiency persist well beyond the nominal end of rush hours.
Varaiya and colleagues have proposed aggressive ramp metering to limit the entry of vehicles to the freeway, so as to keep flow from reaching the critical level that leads to unstable, breakdown traffic flow. But in nearby Orange County, of course, we now have eight years of data on the ability of market pricing to do exactly the same thing. The two eastbound lanes of the 91 Express Lanes, during the afternoon rush hour, typically handle 1,700 vehicles/lane/hour at 65 mph, while the four regular lanes struggle along with about 1,200/lane/hr at about 25 mph. The Express Lanes, with 33% of lane capacity, handle over 40% of the traffic at rush hour. Hence, the attraction of HOT lanes as an alternative way of managing traffic flow.
Some have argued that because HOT lanes improve traffic flow on only a portion of the congested freeway, Varaiya's aggressive ramp metering is preferable to a HOT lanes approach. But the big problem with this kind of really serious, comprehensive metering is that it shifts the congestion problem from the freeway to the on-ramps. Long queues of cars waiting potentially quite long times would back up onto surface streets. And that, in turn, would lead to political pressures to ease off on the metering, probably to the point of minimal effectiveness, at least in comparison to Varaiya's ideal.
But a new alternative has been proposed by Patrick DeCorla-Souza, head of the Value Pricing Pilot Program at the Federal Highway Administration. Since pricing works well to manage traffic flows, why not apply it to all lanes on the freeway, he argues in two new papers presented at the January 2004 Transportation Research Board annual meeting. In his proposed FAIR Network approach, the entire freeway would become a HOT lane, with carpools and transit vehicles going for free while single-occupant vehicles pay a variable toll aimed at maintaining free flow in all lanes. To deal with political opposition to tolling existing lanes, he would use significant portions of the toll revenue to (1) subsidize extensive bus service on the tolled freeways, and (2) provide credits to low-income commuters which they could use either for transit service or toll payments. In addition, toll revenues would be used to improve parallel arterials by adding advanced signal-timing systems.
In one of the two papers, DeCorla-Souza models both a FAIR Network and a HOT Network applied to the Washington, DC region's freeway system. In each case, one additional lane would be added in each direction, along with park-and-ride facilities and special connector ramps. While the FAIR Network requires larger park-and-ride facilities, the HOT Network requires considerably more access ramps and interchange flyovers, to keep premium-toll payers from getting caught in regular-lane traffic jams. And in both cases, the modeling assumes significant subsidies for bus service from toll revenues.
When all the numbers have been crunched, DeColra-Souza finds that the FAIR Network produces more revenue and more benefits than the HOT Network. This is not really surprising, given that the former charges a more modest toll to a much larger number of drivers. Premium tolls charged to a niche market of time-sensitive drivers still produce a lot of revenue, but the much larger number of toll-payers in the former case drives total revenue considerably higher.
Should we conclude from this that current efforts to develop HOT lanes and HOT Networks should be terminated in favor of working to implement FAIR Networks instead? I don't think so, because the FAIR Networks approach is going to be a much tougher sell (and HOT lanes and networks are already a tough sell, though starting to get easier). First, there is the enormous difficulty of persuading drivers and elected officials that paying a toll to use freeways that have always been free and that "we've already paid for" is really in their best interest. This alone, I think, will make FAIR Networks a non-starter in most settings.
Second, the FAIR proposal is a lot more complicated to explain, not only to public officials but also to the media and the public. If you can't summarize it in a sentence or two, it will be a much harder sell. Third, there will be political resistance in some quarters to using large amounts of the toll revenues to subsidize transit and low-income commuters. And fourth, enforcing the HOVs-go-free provision on a six or eight-lane freeway will present major problems, raising the specter of widespread toll evasion.
Peter Samuel (www.tollroadsnews.com) suggests that since adding HOT lanes to four- and six-lane freeways in older metro areas may be all-but impossible, FAIR Networks may be a good fit in those locales, while HOT Networks are a natural for the more expansive freeways of metro areas like Houston and Los Angeles. And the more egalitarian political cultures of, say, Portland and New York may be more comfortable with the FAIR Networks approach than the more individual-choice-oriented cultures of Atlanta and Dallas, where HOT Networks seem to have a bright future.
The bottom line is that we know a lot more about the dynamics of traffic flow than we did 10 years ago, and we know it can and should be managed. All things considered, networks of HOT lanes still look to me like the most practical way forward.
Robert W. Poole Jr. is director of transportation studies and founder of the Reason Foundation.