In December 2018, Madrid introduced a low-emission zone to reduce local air pollution from road transport, in particular nitrogen dioxide concentration. The concentration of nitrogen dioxide exceeded the limit set by the EU Clean Air Programme, which was an average annual maximum of 40 µg/m3. Extended exposure to a high concentration of nitrogen dioxide can cause respiratory, cardiovascular and neurological problems that may lead to illness and premature death. 

The policy restricted access to the city centre for vehicles that did not meet emission requirements according to the efficiency labels of Spain’s national traffic authority. 
 

Policy interventions are often hard to evaluate because it is impossible to know what would have happened if a policy had not been in place. In this study, the alternative scenario is built upon nitrogen dioxide concentrations from a subgroup of meteorological stations within Madrid’s autonomous community where the influence of the policy is deemed negligible. The counterfactual of Madrid-without-a-low-emission-zone (“baseline”) was estimated using the robust synthetic control method. Comparing the baseline scenario and the policy scenario highlights a considerable decrease in nitrogen dioxide concentrations within the Madrid Central low-emission zone. This reduction is represented by the gap between the two lines in figure 1 (below). Starting in January 2019, the line representing the policy scenario (in pink) starts to be noticeably lower than the line representing the baseline scenario (in blue). This gap expanded further in March, when Madrid Central entered into enforcement, after a 3-month grace period. Between January and June 2019, nitrogen dioxide concentrations within the low-emission zone were 27% lower than in the baseline scenario. The gap between the two scenarios peaked in April 2019 with a 43% estimated reduction in nitrogen dioxide emissions.
 

Although the start of the policy brought a substantial reduction in nitrogen dioxide concentrations in the first half of 2019, these gains were short-lived. Starting in July 2019, there was almost no difference between the baseline and policy scenarios. In the July 2019-February 2020 period, the policy scenario only showed a 3% average reduction in nitrogen dioxide concentrations, which is within the model’s margin of error. The study could not explore further impacts of the policy after February 2020 due to the outbreak of the covid-19 pandemic, which brought major changes in mobility and pollution. 

In addition to the impact within the low-emission zone, the policy was also expected to reduce pollution in surrounding areas, as a result of a so-called spillover effect. Out of the four meteorological stations closest to Madrid Central, the only one which presented a substantial reduction of nitrogen dioxide concentrations is that located in the Retiro Park. At this station, between December 2018 and September 2019, there was an average reduction of 18% compared to the baseline scenario. However, data from the three other stations showed no effect from the implementation of the policy. Méndez Álvaro station only experienced reductions in the first two months after Madrid Central was implemented, while no effect can be identified for either Escuelas Aguirre or Plaza de España stations. These two stations also present concentration levels substantially above the 40 µg/m3 limit set by the EU. The lack of an effect in Plaza de España is especially surprising, since this station is located at the edge of the low-emission zone, on a busy crossroad. 
 

These results show that the potential for reducing nitrogen dioxide concentrations in the surrounding areas of the low-emission zone was limited, and the policy left several areas with concentrations that substantially exceeded EU maximums. This finding suggests that low-emission zones should be extended to cover all areas where pollution needs to be reduced, without relying on spillover effects. 

Looking at the number of violations of the low-emission policy provides a possible explanation for the policy’s sudden drop in effectiveness. After the grace period, monthly violations (as registered automatically by camera) between March and May 2019 averaged below 10,000 per month. This period showed the largest reductions in nitrogen dioxide concentrations. However, starting in June 2019, the number of violations climbed sharply to around 80,000 per month during the summer of 2019 and over 100,000 per month between September and November 2019. After November, the figures gradually fell but remained substantially higher than during the period immediately after implementation.

The data suggested that drivers stopped complying with the policy in June 2019. This might have been caused by the change in Madrid’s new municipal administration, which took office that same month. The winning party had campaigned against the Madrid Central low-emission zone, which was a flagship programme of the outgoing municipal government. As a matter of fact, the new administration tried to suspend the policy the month it took office, but the attempt was ultimately blocked by a local court. This manoeuvre could have generated uncertainty among drivers regarding the ongoing enforcement of the policy. This study highlights how uncertainty and/or a loss of credibility regarding a regulatory measure might substantially affect its effectiveness. Nonetheless, results show that the low-emission zone seems to have reduced nitrogen dioxide concentrations drastically during the first months of its implementation.

ABADIE, A. (2021): "Using Synthetic Controls: Feasibility, Data Requirements, and Methodological Aspects", Journal of Economic Literature, 59(2), 391-425. 

AMJAD, M., D. SHAH y D. SHEN (2018): «Robust Synthetic Control», Journal of Machine Learning Research, 19, 1-51