This paper aims to correct standard measures of agglomeration economies to account for air pollution generated by commuting. This paper examines the impact of nitrogen oxide (NOX) a pollutant mainly released by transportation on worker productivity. On one side, literature on agglomeration economies highlights the positive role of employment density on productivity, without accounting for the environmental impact of a better accessibility. On the other side, several studies (Graham, 2007; Rice et al., 2006) show that new transportation infrastructures or policies have a positive effect on accessibility, thus enlarging the opportunities offered to workers and leading to increased labor productivity. However, additional commuting trips may be induced by enhanced accessibility, and generates higher levels of polluting emissions (Goodwin, 1999; Litman, 2011). Epidemiological studies show that atmospheric pollution has a negative and significant impact on human health (see e.g., Currie et al. (2009a, 2009b)), implying lower labor productivity (Lavy et al., 2012; Graff Zivin and Neidell, 2012). This article aims at correcting estimations of agglomeration economies by accounting for air pollution resulting from commuting. More specifically, we consider local air pollution as a determinant of labor productivity and integrate it in the theoretical framework estimating agglomeration economies in order to assess the extent to which pollution limits the full efficiency of production capacities. We use aggregate data for the year 2009 for the 304 French metropolitan employment areas. First, we estimate the effects on labor productivity per worker of employment density, controlling for standard variables. In line with the literature, the results show an increase in productivity of 0.03% for a 1% increase in employment density. Second, we include NOX emissions as a proxy for atmospheric pollution. In line with epidemiological studies, we find that air pollution negatively impacts labor productivity. A 1% increase in the level of NOX emissions leads to almost 0.1% decrease in productivity. Third, we compare the models with and without air pollution. When pollution is accounted for, the positive effect of employment density on productivity is reduced. Finally, we focus on an illustrative case to show the magnitude of the reduction of agglomeration economies when local air pollution is considered. When NOX emissions are included in the model, the productivity gains of agglomeration are reduced by more than 13%. We conclude that accounting for air pollution (NOX) in the econometric model of agglomeration economies allows 'correcting' the effect of density on productivity, confirming that pollution matters. Indeed, introducing new transportation infrastructures and policies improves the accessibility of an area, which enlarges the positive externalities (agglomeration economies) stemming from increased density, but which also generates additional pollution due to induced trips. Yet pollution limits the full efficiency of production capacities.
