For several decades, using variable pricing to address freeway congestion has been a goal of transportation planners. Not only would such pricing reduce congestion (by addressing the imbalance between capacity and rush-hour demand), but it would also generate useful information about where new capacity was most needed, and provide funding to pay for that capacity.
Yet despite federal efforts that began in the 1970s and have continued to the present, no urban area has been enticed into putting congestion pricing onto its freeway system. Doing so is considered politically impossible, primarily due to the opposition of highway users—motorists and truckers. These groups argue that paying tolls to use existing freeways would amount to paying twice (or "double taxation"). Some doubt that pricing would work; others that it would work so well that hardly anyone could afford to drive. And since nearly every adult of driving age is a motorist, it's easy for elected officials to equate "motorist" with "voter."
Even in huge metro areas overseas, congestion pricing has been implemented successfully only three places in three decades: Singapore, London, and Stockholm. It has been defeated in Hong Kong, Kuala Lumpur, The Netherlands, and Manchester, among others. These overseas metro areas have much lower car ownership and much larger transit systems than typical large U.S. metro areas. This suggests that even where alternatives to driving are more viable, congestion pricing is still very tough to sell.
The underlying problem is that congestion pricing can easily produce more losers than winners. Those who pay the toll would be winners, because their time savings would be worth what they had to pay. But a great many others would be “tolled off” the freeways, to already-congested arterials. And those using the arterials would be made even worse off by those diverting to them from the tolled freeways.
To solve this problem, we need to think outside the box. All conventional freeway pricing proposals make two unexamined assumptions: that everyone would be charged the same price and that all lanes would serve all vehicles. Let's re-examine those assumptions.
Detailed research on the congestion-priced HOT lanes on SR 91 in California by UC-Irvine economist Kenneth Small finds a very large distribution of values of time and values of trip-time reliability among all motorists in that congested corridor. Small and colleagues modeled several pricing alternatives for that corridor, including the existing combination of priced and free lanes, the standard all-lanes-priced model of freeway pricing, and a dual-price model. The last of these produced the highest social welfare outcome, because it provided more total congestion relief without "tolling off" large numbers of users. Stephen Shmanske of Cal State Hayward (East Bay) first proposed this approach in 1991, and modeled a dual-price system for the Golden Gate Bridge in 1992.
The other unexamined assumption that needs a fresh look is that all lanes should be configured as "general purpose" (GP) lanes. Highway engineers know that throughput is somewhat higher on two GP lanes than on two parallel, but separated, lanes. That's because with multiple lanes, faster vehicles can pass slower ones. And transportation economists can point out that because of the "lumpiness" of traffic lanes, it's difficult to provide the right amount of lane capacity restricted to a single use—say buses, HOVs, or trucks. Most such specialized lanes are under-utilized.
But we now have more than 15 years experience with HOT lanes, most of them single-lane-per direction. Thanks to variable (congestion) pricing, these lanes can be filled to a target level of traffic (e.g., Level of Service C) and kept that way. In selected corridors, there may be enough truck traffic to justify truck-only lanes, and several such projects are under detailed study around the country.
My proposal for freeway pricing draws on these points to suggest that transportation planners should revise their goal. Instead of one-price-fits-all on GP lanes, we should be aiming at a three-part system: premium lanes (with premium pricing) for cars, light delivery vehicles, and commuter buses; regular lanes with modest peak-period pricing; and in selected corridors, truck-only lanes. This approach should produce more winners than losers and therefore be politically feasible.
Moreover, it can be phased in over time, beginning with a network of premium-priced lanes, as currently planned for Atlanta, Dallas, Houston, San Diego, San Francisco, and Seattle. Only after that network is operational should planners suggest the added benefits of modest peak-only pricing on the remaining GP lanes.
Freeway congestion pricing is still a worthwhile goal. But we are more likely to reach it by working harder on the details.