Find the full-length report draft here.
New York’s political community and the general public have yet to come to terms with reality on congestion pricing. While COVID-19 has suppressed travel demand across the region — deeply for now and to an uncertain extent over the next several years — that decrease has been concentrated in mass transit. Bridge and tunnel crossings into Manhattan are, as of this summer, only down 9% from the pre-COVID baseline. The average daytime travel speed in Manhattan below 60th Street is already back down to 8MPH, barely above the pre-COVID average of 6.9MPH. As the recovery continues, traffic will only get worse — and we need a flexible and dynamic tolling regime that can “roll with the punches” of varying traffic and economic fluctuations as it permanently solves the congestion problem.
The paper offers four novel suggestions for congestion pricing in New York City:
1. A speed target should be at the center of policy, rather than revenue, and the fee should vary dynamically in response to traffic volumes to achieve the target speed subject to a maximum peak toll. (Subject to political constraints, the closer the peak toll cap can get to $26, the better the economically estimated balance of speed and toll price).
2. Upstream tolls should be credited broadly to increase regional equity and ensure the incentives to choose any given route to Manhattan depend only on traffic management, not on revenue considerations.
3. Dynamic tolls on each entry point to Manhattan should float independently to incentivize only useful “toll shopping”. Current toll differences on priced and unpriced crossings are arbitrary and divert traffic to the busiest free crossings, while independently floating tolls would equilibrate to balance traffic volumes across all crossings.
4. The cap on licenses for For-Hire Vehicles should be removed and the fixed pickup tax should be replaced by a dynamic per-mile toll tailored directly to the per-mile congestion externality. While distance-based charging inside the cordon would work better for all vehicles in theory, in practice the privacy and political challenges of vehicle tracking have only yet been solved for taxis and FHVs.
The two key things to understand are:
- The existing MTA plan to implement “scheduled variable tolls” in Manhattan does not set tolls high enough to eliminate congestion in Manhattan, and does not include a policy speed target.
- Any variable toll scheme that ensures a minimum travel speed above 10MPH at all times in Manhattan would raise up to $5 billion annually, which would help the MTA financially bridge the COVID emergency and eventually pay off its excessive debt.
All else equal, traffic speed is a function of traffic volume. At higher volumes, speed declines. The relationship between speed and volume is called the “speed-volume curve”. Every day we observe this relationship: We can observe how many cars enter or travel within the Manhattan grid during a given hour and measure average traffic speeds during that time. At night, traffic moves at the speed limit. During the worst of rush hour, it moves below 5MPH.
Exploiting this simple empirical relationship to derive an average speed-volume relationship for the Manhattan grid, we can show the true average hourly traffic capacity of the grid. For example, the Manhattan grid can handle 91,000 hourly vehicle-miles (VMT) of travel at 20MPH, or 143,000 hourly VMT at 10MPH.
Figure 1: Traffic Volumes Consistent with 10MPH or 20MPH in Manhattan
If we want Manhattan traffic to move at 10MPH, hourly VMT can’t exceed 143,000. If we want traffic to move at 20MPH, hourly VMT can’t exceed 91,000. Any tolling scheme intending to get traffic moving needs to achieve volume reductions corresponding to the intended speed benefits. This is not a political thing or an economic thing, it is a geometric relation of vehicle volume to vehicle speed in the space available in Manhattan and its bridges and tunnels.
The optimal speed and toll is not a mere political question, even though it is likely to be chosen by political expediency. There is an actual fact of the matter: By observing the travel time losses imposed by each incremental mile traveled in real time, and multiplying that time loss by the average value of travel time based on wages in the NY metro area, we can give a precise estimate of the un-priced “congestion externality” imposed by an incremental mile of vehicle travel in varying conditions. Consider the full speed-volume relation chart with an imputed pro forma value of travel time at $30/hour:
Figure 2: Manhattan’s Speed-Volume Curve in 1 MPH Increments:
[Note: 6.9MPH is highlighted because it was the average daytime travel speed in the Manhattan CBD pre-COVID. The July 2020 daytime average was 8MPH, consistent with a ~$10/mile average daytime congestion externality for passenger vehicles on the Manhattan grid]
This chart shows how adding one more vehicle-mile of travel to the Manhattan grid slows down all other simultaneous vehicles, and translates that travel time loss into dollars at $30/hour. In the full paper draft, I translate the data behind this basic schedule of the “traffic volume-contingent congestion externality” on the Manhattan grid into a dynamically variable tolling system that permanently eliminates excessive congestion in Manhattan while raising up to $5 billion annually (in the pre-COVID baseline). The key is dynamic variable tolling with a speed target, where the one-way toll varies in short increments as necessary, subject to a maximum toll cap of $26, to suppress traffic volumes to achieve a policy target speed.
This paper relies on the Hayekian logic of dynamic pricing to build in robustness to error in our expectation of the responsiveness of traffic to price — much more so than a fixed toll schedule that policymakers would scramble ex post to change in the event of an economic depression or other unexpected negative exogenous shock to traffic volumes. Intellectual humility is warranted insofar as the headline speed target is subject to some model uncertainty. (It’s especially useful while facing uncertainty in the upside risk of excessive traffic volume recovery post-COVID.)
Specifically, the speed-volume curve tells us exactly how much to charge the next car to travel a mile for any given number of hourly vehicle-miles of travel in Manhattan. For any given volume of travel, we know how much the next vehicle mile will slow down all the other cars already on the road, and can then reliably convert those minutes of aggregate travel time loss into a dollar value of delay per mile without knowing anything about human behavior in response to tolls. We know, for example, that when hourly traffic volumes hit the PM peak of a little over ~170,000 VMT in Manhattan, corresponding to 6MPH on average, then the next average cordon entry causes about ~$26 in slowdown to all the other cars combined. That’s why this paper sets the ideal maximum toll cap at $26. Such a toll is “optimal” in the sense that trips valued in excess of the delay cost should keep happening, while trips valued less than that should not.
But federal legislation allowing dynamic tolling lanes on federally funded Interstates requires the additional step of setting a concrete minimum policy speed target, not merely charging the abstract volume-conditional Pigouvian toll schedule. So we know that roughly $26 is the maximum correct price to charge per entry, according to the speed-volume curve and our knowledge of the average value of travel time, when traffic is at its average hourly worst in the pre-pandemic baseline during the PM rush hour. But we must then translate the known $26 per-entry toll from the speed-volume curve into an estimated speed target achieved through a cordon toll. To do this the BTA model has to commit to an estimate of travel time and price elasticities and average travel distance per cordon entry by different vehicle and trip types — and the BTA happens to expect the human response to the $26 peak rush hour toll to coincide with roughly a 10MPH speed target. The speed-volume curve is a geometric fact, and the $26 toll cap is readily derived from the local value of travel time, but the next step, of projecting of human behavior in response to the toll equilibrating at 10MPH is an out-of-sample estimate.
To reiterate, we can be confident from the baseline speed-volume curve that $26 is the maximum necessary one-way toll by consulting the speed-volume relation chart corresponding with the worst average hourly congestion, but whether that optimum coincides with the 10MPH rush hour speed target in equilibrium relies upon the empirical accuracy of the BTA model. Less-responsive traffic would result in the $26 peak toll lasting longer throughout the day than currently projected. More responsive traffic could yield a higher optimal speed target than currently projected to result from the $26 optimal cap.
This empirical uncertainty favors a more dynamic form of variable tolling — I argue that true dynamic variable tolling is appropriate for Manhattan, but Singapore’s hybrid approach of quarterly adjustments to scheduled variable tolls to achieve an average speed target over a longer time horizon would also represent a realistic second-best option.
The key above all else is to target speed, not revenue, in order to ensure the traffic speed target is achieved without excessive tolling in an error-proof, continuously updating fashion.
To read more, find the full-length report draft here.