Impact of Low-Emission Zones on Spatial and Economic Inequalities using a Dynamic Transport Simulator

André de Palma & Lucas Javaudin

THEMA, CY Cergy Paris Université

PSE, March 2024

Introduction

Context

  • Road transport sector is responsible for 37 % of nitrogen oxides emissions (NOx) in Europe (EEA, 2021)
  • Nitrogen oxide pollution causes around 40,000 premature deaths yearly in Europe (EEA, 2021)
  • Air pollution causes about 7,920 premature deaths yearly in Paris' urban area, Île-de-France (AirParif, 2022)
  • Popular instrument to improve air quality: Low Emission Zones (LEZ)

Low Emission Zones

  • Low Emission Zone: area in the city center where the most polluting vehicles cannot travel
  • In Europe, LEZs have been implemented in hundreds of cities as of today
  • In France, 25 cities have implemented LEZs; cities are forced to implement a LEZ when pollution is above a threshold level

Crit'Air System

  • Entry restrictions in the LEZ are based on the Crit'Air system
  • Each vehicle is put in a Crit'Air category based on its fuel type (diesel, petrol, electric, etc.) and its age

A Controversial Policy

Arguments in favor of LEZ
  • Improved air quality
  • Decreased CO2 emissions
  • Less noise pollution
  • Less congestion
Arguments against LEZ
  • Restriction of freedom
  • Unfair: penalize people stuck with an old vehicle (poor and rural households)
  • Some drivers might make a detour around the LEZ, polluting even more
  • Drivers are induced to purchase new vehicles before their end-of-life

Paris' Low Emission Zone

  • Paris and 76 neighbor municipalities
  • 367 km2 area (3 % of Île-de-France)
  • 5 M inhabitants (40 % of Île-de-France)
  • A86 highway enables detours around the LEZ
  • Since January 2025: Vehicles Crit'Air 3 or worst are banned
  • 68 € fine for non-respect

Île-de-France Vehicle Fleet

  • Municipality-level vehicle fleet data (with Crit'Air categories) from the Ministry of Ecology
  • Extrapolation to predict the fleet in 2025
  • In 2025, around 21 % of vehicles in the region would be Crit'Air 3 or worst

Research Questions

Analysis of Paris' LEZ with a transport simulator (METROPOLIS2)
  • What is the global impact of the LEZ (on air quality, congestion, etc.)?
  • Which people are winning and losing from the LEZ?

Methodology

Introduction

  • Scope:
    • Île-de-France
    • Trips for an average working day
    • Five modes : car (driver), car (passenger), public transit, bicycle and walking
    • All trip purposes

METROPOLIS2

  • METROPOLIS2 is an agent-based dynamic mesoscopic transport simulator
  • Simulation of mode, departure time and route choice
  • Congestion simulated from bottlenecks with queue propagation (spillback)
  • Computation of pollutant emissions and exposure of population to pollutants with the METRO-TRACE module

Input Data

  • Road network: OpenStreetMap [72,962 km of roads]
  • Public transit timetable: GTFS [1850 lines]
  • Trips : synthetic population (Hörl and Balac, 2021) [2,451,841 persons ; 8,774,929 trips]

Choice Models

  • Mode choice: multinomial Logit model
  • Departure time choice: continuous Logit model
  • Route choice: fastest path
  • Surplus computed from the log-sum formula

Utility function

Utility of trip $k$ with mode $j$:
  • $t^{\text{d}}$: departure time ; $t^{\text{a}}$: arrival time
  • $c_j$: mode constant ; $m_{k | j}$: monetary cost (fuel)
  • $\alpha_j$: value of time
  • $\beta_k$: early-arrival penalty ; $\gamma_k$: late-arrival penalty ; $t^*_k$: desired arrival time

Additional Hypothesis

  • Car (driver) option only to car owners with driving license
  • Car (passenger) option only available to car owners
  • Public transit travel time computed with OpenTripPlanner and GTFS data
  • Bicycle speed: 10 km/h
  • Walking speed: 4 km/h
  • 600 k truck trips are simulated
Mode Availability (baseline) Availability (LEZ) Constraint
Car (driver) 53 % 49 % Authorized car + driving license
Car (passenger) 74 % 68 % Authorized car
Public transit 54 % 54 % Feasible itinerary
Bicycle 100 % 100 %
Walking 100 % 100 %

Calibration

Step Calibrated parameters Target values Target source Methodology
1 Road constant penalties and free-flow speed Free-flow travel times TomTom API LASSO regression
2 Road capacities Time-dependent congested travel times TomTom API OLS regression
3 Schedule-delay penalties by purpose Distribution of departure times by cluster Travel survey (EGT) Bayesian Optimization and Gaussian Process
4 Mode-specific utility parameters by socio-demographic characteristics Mode shares by cluster Travel survey (EGT) Random Forest regression

LEZ Policy Evaluation

  • Two METROPOLIS2 simulations:
    • Baseline simulation (calibrated): no LEZ
    • LEZ simulation (counterfactual): January 2025 LEZ (Crit'Air 3 and worse)
  • Limits:
    • Short-run analysis: no car-ownership model, no relocation (of activities or homes)
    • Temporal restrictions of the LEZ not considered
    • Exceptions and cheating not considered

Results

Aggregate Results

Baseline LEZ Variation
Average travel surplus -28.96 € -29.03 € -0.07 €
Average travel time 01:12:05 01:12:55 +50s
Vehicle-kilometers 123.30 M km 119.72 M km -2.9 %
Mode shifts
Baseline / LEZ Car (driver) Car (passenger) Public transit Bicycle Walking Total
Car (driver) 28.6% · 1.4% 0.2% 0.3% 30.5%
Car (passenger) · 5.3% 0.2% · · 5.5%
Public transit 0.4% 0.1% 18.4% · · 18.9%
Bicycle · · · 1.1% · 1.1%
Walking · · · · 43.9% 43.9%
Total 29.1% 5.4% 19.9% 1.4% 44.2% 100%

Congestion Impact

  • Congestion decreases on the main highways inside the LEZ (Boulevard Périphérique and A1 motorway)
  • Little impact outside the LEZ

Public Transit Flows Impact

  • Larger flows on most legs, mainly in the surroundings of Paris (North, East and South)
  • RER A: +1.2% passengers-kilometers
  • RER B: +2.1% passengers-kilometers
  • Tramway T7: +24.4% passengers-kilometers

Pollutant Emissions

  • Emissions of PM2.5 and NOx generated by road traffic are computed from the EMISENS model
  • Emissions depend on vehicles fuel type and age as well as instantaneous speed (link-level)
  • Emissions decrease more inside the LEZ
Baseline LEZ Variation
PM2.5 emissions 2.83 tons 2.66 tons -6.0 %
NOx emissions 33.32 tons 30.45 tons -8.6 %
CO2 emissions 21 730 tons 20 829 tons -4.1 %

Pollutant Dispersion

  • PM2.5 and NOx emissions are dispersed in the air using a Gaussian Plume model
  • Dispersion depends on wind's average speed and direction (Eastward, 10 km/h)
  • Concentration in pollutants decreases more inside the LEZ and East of the LEZ

Population Exposure to Pollution

  • Health impact is a function of the increase in mortality due to exposure to pollutants, given the concentration levels
  • Exposure is computed based on the actual location of individuals in time and space
  • Exposure decreases more near Paris (high concentration and high population density)
Baseline LEZ Variation
PM2.5 premature deaths 5.9 5.3 -9.4 %
NOx premature deaths 5.4 4.9 -10.1 %
Health surplus -12.537 M € -11.312 M € -9.8 %

Health Impact: Summary

Variation
Vehicles-kilometers -2.9 %
PM2.5 emissions -6.0 %
PM2.5 premature deaths -9.4 %
Analysis in vehicle-kilometers is not precise enough:
  • The decrease in vehicle-kilometers is stronger for the most polluting vehicles
  • Air quality mostly improves inside the LEZ where population density is higher

Heterogeneous Impacts

  • Health impact: between 0 et +30 cents per day per individual
  • Travel impact:
    • 72 % are not significantly impacted (variation smaller than 1 cent)
    • 3.5 % "win" more than 1 € daily
    • 3.3 % "lose" more than 1 € daily

Winners and Losers Characteristics

Winners Losers Others
Criteria $\Delta \text{surplus} \geq 1$ € / day $\Delta \text{surplus} \leq -1$ € / day $|\Delta \text{surplus}| \lt 1$ € / day
Average travel surplus variation +2.0 € -7.2 € +0.0 €
Authorized car owners 97 % 4 % 53 %
Banned car owners 3 % 96 % 19 %
Mean trip distance 45 km 32 km 15 km
Access to public transit 51 % 64 % 85 %
Car mode share (baseline) 79 % 79 % 33 %
Car mode share (LEZ) 83 % 1 % 33 %

Winners and Losers Location

  • "Winners" are spread over the region
  • "Losers" are mainly living along the LEZ
Winners
Losers

Winners and Losers Income

  • Share of "winners" slightly increasing with the municipality average income
  • Share of "losers" uncorrelated with the municipality average income
Winners
Losers

Conclusion

Conclusion

  • Methodology for the evaluation of public policies with a transport simulator
  • Global impact: decrease of car use, vehicle kilometers, congestion and pollution
  • Individual impact:
    • Health impact distributed evenly across the population
    • Travel surplus impact shows great disparities
  • Characteristics of the winners and losers of the policy
  • Limits:
    • No analysis of the income effect at the individual level
    • Short-run analysis: no car-ownership model, no activity-based model, no location choice model

Thank you