Tolls vs tradable permits for managing travel on a bimodal congested network with variable capacities and demands

Highlights

• Tradable Permit Schemes (TPS) are a promising tool to control traffic congestion.

• We compare the efficiency of TPS and tolls on a bimodal network under uncertainty.

• Neither the toll nor the permit quota can depend on demand or link capacities.

• The toll is invariant to some model parameters, whereas the permit quota is not.

• The toll also outperforms the TPS numerically in most instances examined.

Abstract

Congestion pricing has long been considered an efficient tool for tackling road traffic congestion, but tolls are generally unpopular. Interest is growing in tradable permits as an alternative. Tolls and tradable permits are interchangeable if travel conditions are unchanging, but not if conditions vary and tolls and permit quantities are inflexible and cannot be adapted to current conditions. We compare the allocative efficiency of tolls and tradable permits on a bimodal network under uncertainty. Road links are congestion prone and public transit may be crowded. Road traffic entering a cordon area around the downtown is controlled using either a toll or a tradable permit. Two groups of travelers can drive or take transit. Group 1 travels downtown, and if it drives it must either pay the toll or use a permit. Group 2 travels to a suburb, and can avoid the cordon by taking a bypass. All demand and cost parameters of the model can vary, either systematically or irregularly. Each parameter combination constitutes a state. Travelers learn the state in advance, and adapt their mode and route choices accordingly. A planner minimizes expected total travel costs by either setting the level of the toll or choosing the quota of permits to distribute. Two cases are considered. In the first, the toll and quota are flexible and can be adjusted to daily travel conditions. In the second, the instruments are inflexible and must be set at the same level regardless of the state. If travelers have identical preferences, the optimal flexible toll is invariant to the numbers of travelers in each group and the capacity of the link entering the cordon. The toll is robust in the sense that inflexibility causes no welfare loss if these parameters vary. By contrast, the quota is not robust.

We derive a general rule for ranking the efficiency of a fixed toll and fixed quota. We then explore a numerical example. In most instances, the fixed toll outperforms the fixed quota by a significant margin although the quota can do better in some states. We also compare the welfare-distributional effects of tolls and permits, and find that suburban travelers fare better than downtown travelers from both forms of regulation.

Introduction

Traffic congestion is a burden worldwide. Inrix reports that in 2019, US drivers lost nearly 100 h to traffic congestion on average, at a personal cost of $1377, amounting to a loss of nearly $88 billion to the US economy.¹ Congestion imposes similar costs in the UK, France, and Germany.² Congestion pricing has long been considered an efficient, if not the most efficient, tool for tackling traffic congestion. Advances in tolling and IT technology have reduced the costs of imposing tolls, billing motorists, and informing them about where, when, and how much they will pay. Yet, only a few city-scale congestion pricing schemes exist: in Singapore (1975), London (2003), Stockholm (2006), Milan (2008), and Gothenburg (2013). Smaller schemes have been established in Durham (2002) and Valletta (2007), and several cities including New York City have plans to introduce charges. However, many proposed schemes have failed. Public opposition and equity concerns are the most frequently cited explanations (Jaensirisak et al., 2005, Grisolía et al., 2015, Krabbenborg et al., 2020).

Due in part to opposition to tolls, quantity controls and regulations are much more widely used than tolls. These include limits on vehicle registrations, license plate number restrictions on daily use, perimeter control of traffic, driving bans, and traffic calming. These measures are often intended to curb not only congestion, but also reduce pollution, enhance safety, and reduce noise. For example, Fageda et al. (2020) report that low emission zones have been implemented in 46 out of 130 large cities in 12 European countries.

Quantity controls are generally less efficient than tolls because they do not allocate road space to those who value it the most. Tradable Permits Schemes (TPS) are a hybrid measure that combines price and quantity controls. TPS can improve on pure quantity controls by granting driving rights, and allowing agents to trade them. TPS have been used to reduce acid rain, lead, and carbon emissions; to administer Corporate Average Fuel Economy (CAFE) standards; and to allocate taxi licenses, vehicle quotas, and airport landing slots. TPS have not yet been applied to road travel. However, advances in information technology and familiarity with it are making TPS increasingly feasible. The idea is conceptually straightforward. Drivers would need a permit to make a trip, access a road, or enter a restricted area. The amount of travel would be regulated by limiting the number of permits that are issued. If permits are distributed free of charge, individuals in aggregate would not incur an additional monetary cost to travel. Furthermore, if permits are distributed on an equal-per-capita basis, lower-income households, which tend to travel less by car, would receive more permits than they use, and would earn income from selling the excess. Vertical equity could be improved further by giving lower-income households more permits.

If travel conditions are known and unvarying, a TPS can be designed to support the same travel pattern as tolls. The two instruments are then allocatively equivalent, with the TPS having a likely advantage in terms of acceptability. However, as initially shown by Weitzman (1974) in a planned-economy context, price and quantity instruments are not equivalent in a nonstationary environment if they cannot be adjusted as conditions change. Moreover, road travel conditions do fluctuate. Travel demand varies predictably by time of day, day of week, and season. Unanticipated fluctuations in capacity and demand are also common. According to US FHWA (2020), about half of total congestion delays arises from nonrecurring events such as crashes, debris, work zones, traffic signal failures, bad weather, special events, and so on. These shocks can cause substantial delays. Closure of one lane of a three-lane highway due to an incident reduces total capacity by about half.³ Heavy rain can reduce freeway capacity by 10%–15% (van Lint et al., 2000, Chung et al., 2006, Brilon et al., 2008), and snow storms and other severe weather conditions by more than 20%. Capacity can also drop if individual drivers change lanes or brake sharply, and cause traffic flow to break down (Brilon et al., 2008). Public transit service disruptions due to strikes and other shocks are also common in some cities, and they can affect traffic congestion if transit users switch to driving.⁴ Rain, snowfall, and high temperatures encourage commuters to drive rather than take public transit or walking (Belloc et al., 2022). Finally, great shocks such as the COVID pandemic can affect travel demand for extended periods (de Palma et al., 2022).

Tradable permits and tolling schemes would still be allocatively equivalent if they could adapt to all these fluctuations. Yet, with the exception of dynamic pricing on some High Occupancy Toll (HOT) lanes in the US and Israel, real-time, state-dependent road pricing has not been adopted (Lombardi et al., 2021). Several explanations have been suggested. The infrastructure, operating, and accounting costs may be too high to justify state-dependent pricing over short time horizons. People may prefer fixed charges because adapting to frequent price changes is difficult (Bonsall et al., 2007), or because the outlays are easier to predict (Li and Hensher, 2010). People may also dislike price uncertainty per se (Lindsey, 2011), and they may dislike having to pay high tolls if they have no good alternative.

The potential for an adaptable TPS is still unknown since no schemes are operational yet. Nevertheless, transactions costs militate against frequently changing either the number of permits distributed, or the number of permit units required to use particular roads or enter certain areas. Permit quotas are likely to be distributed weekly, monthly, or quarterly rather than daily, and requirements are likely to be set systematically (e.g., by day of week or season) rather than adjusted according to current travel conditions.⁵

In this paper, we assume that tolls and tradable permits allocations are inflexible, and thus cannot be adjusted on the basis of daily travel conditions. We compare the allocative efficiency of tolls and permits in the face of shocks to demand, road capacity, and transit, assuming no administration, transactions, or traveler compliance costs are incurred for either instrument. Unlike earlier studies that compare tolls and permits under variable conditions, in our model the instruments are applied over only part of the transport network. Hence, they are second-best and cannot support a first-best optimum even under stationary conditions. We summarize our main results at the end of the literature review in the next section.