Many studies have underlined the significant threat that air pollution poses to human health and wellbeing: proofs of the connection between health damages (lung cancer, mental wellbeing, cardiovascular diseases) and air pollution have multiplicated in recent years, but knowledge of dangers due to human exposure to air pollutants dates back at least to the London “great smog” in 1952, that caused 12,000 deaths. Recognizing this challenge, the World Health Organization (WHO), the European Commission, and several international bodies have put forth policy strategies to mitigate air pollution.

This study provides much-needed evidence of the determinants of air pollution, considering simultaneously the several gases that compose air pollution: Particulate Matter 10 and Particulate Matter 2.5, Sulphate Dioxide, Nitrogen Dioxide, Carbon Monoxide, and ground-level Ozone concentrations.

Current models for assessing air quality are often not adequate since they consider pollutants in isolation or use simplistic aggregation techniques, neglecting the complex interactions among diverse air quality variables. In response to these limitations, there is a pressing need for a more reliable and comprehensive air quality index that considers the complexity of the phenomenon.

The aim of this paper is to propose a statistical approach to develop an air quality index that overcomes the drawbacks of existing models. Leveraging Structural Equation Models (SEMs), a methodology widely used in many fields of research, but less explored in environmental research, the first innovation proposed in this study introduces a novel model for the six pollutants and constructs a model-based multidimensional Air Pollution Index (API) that takes into account the complex relationships among pollutants.

This is called API_2 and can be used to monitor the quality of the atmosphere. The new index ranges in 0-1 and the closer it is to 1, the more polluted the air of the city is. 

The critical importance of implementing rigorous measures to reduce human exposure to air pollutants needs studies that identify the drivers of air pollution. Focusing on 130 metropolitan areas of the European Union, this study provides estimates of such determinants, the most impactful of which  is the number of cars circulating in urban areas. The analysis presented in this paper also identifies Northern European cities as the less likely to be polluted. Socio-economic features of metropolitan areas relate to the comprehensive level of air pollution: “younger” cities are less polluted (the youth dependency ratio presents a negative coefficient) and a negative effect on the air pollution level is also associated to a higher level of participation to the labour market.

This paper proposes an original statistical procedure that can be applied to derive indexes measuring different phenomena by considering the optimal specification of a SEM, a powerful statistical tool that enables a comprehensive understanding of relationships within complex systems. According to this procedure, each step of the index construction process relies on statistical choices made considering the goodness of fit of the model with the observed data as well as the statistical significance of the obtained results.