A study of nearly 10,000 Israeli drivers finds that the economic case for congestion charges depends on who can change their travel plans and what happens to the money collected, not simply on whether traffic becomes lighter.
The October research combines a driving experiment with a model of the road network. It separates the behavior observed among participants from estimates of what a wider charging system might achieve.
That approach addresses a familiar objection to road tolls. Faster journeys can benefit drivers and businesses, but people who cannot change their routes or working hours still have to pay.
Itai Ater, Brad Ross, Adi Shany and Shoshana Vasserman explain the analysis in Road Pricing under Heterogeneity, an October NBER working paper. It has not been peer-reviewed. An author-hosted version dated October 6 contains the full experiment and modeling details.
What is congestion pricing?
Congestion pricing charges for road use in ways intended to reduce demand at crowded times or places. It can involve a fee to enter an area, a charge per distance driven, or a toll that changes with the time of day.
The US Federal Highway Administration’s road-pricing guidance distinguishes these arrangements from flat tolls and describes dynamic charges that change with conditions. Drivers need alternatives they can use for prices to change their choices at short notice.
A charge is a direct cost to the person paying it. The intended benefit is that some journeys move to a different time, route or transport mode, reducing delays for others. Assessing the policy requires counting both sides.
The experiment changed the cost of a journey
The Israeli experiment ran from November 2019 to November 2021. Drivers were initially monitored for six months without financial incentives, then faced time- and location-dependent per-kilometer charges deducted from a virtual budget. The remaining budget was paid to them in cash.
Participants consequently had a reason to avoid costly journeys, although the arrangement was not an ordinary toll bill paid from their own bank accounts.
The authors used staggered recruitment to compare drivers who had started facing charges with participants still in the monitoring stage. They found reductions in driving at congested times and places, through changes including trip frequency and departure times.
Responses were uneven. Drivers with more flexible schedules and those with better relevant public-transport alternatives were more able to adjust.
That difference can matter to employers. Allowing a shift in starting time may give an employee an option unavailable to someone who must be at a fixed workplace for a fixed shift. Our reporting on remote work and commuting distances examines another way workplace attendance can affect travel choices.
Revenue can change who gains
The researchers then modeled a charging system across the Central District around Tel Aviv. Their simulations account for overlapping routes: removing a bottleneck in one place can alter traffic farther along the network.
With full recycling of toll revenue, the experiment’s price schedule produced an estimated economic-welfare gain comparable to adding a lane to every highway in the modeled area, before construction costs. Economic welfare here is the model’s measure of benefits and costs, not an observed increase in company profits or household income.
The model also generated enough revenue to compensate those made worse off. That is a finding about what the modeled policy could finance, not evidence that compensation was paid under a real nationwide charging scheme.
Returning money to drivers or spending it on services they value changes the net calculation. A faster journey and a toll payment cannot be judged in isolation from the use of that payment.
Results depend on the network and the people using it
The participants volunteered, and the experiment covered a period that included the pandemic. Their behavior is not a direct forecast for every city’s drivers or for current traffic conditions.
The wider gains depend on assumptions about travel preferences, road capacity and alternatives. The model does not fully capture long-term changes in where people live and work.
For a business running vehicles, a shorter journey may offset part of a toll, while a worker with little scheduling freedom may face a new expense with limited room to respond. A city’s assessment needs to show those different outcomes, alongside a specific plan for the revenue.
