Bringing down levels of carbon monoxide and particulate matter in the air of Mexico City by just 1% each would result in a reduction of approximately 21 infant deaths per year. That is one of findings of a new study of the short-term impact of exposure of young children to air pollution.

The research by Eva Arceo, Rema Hanna and Paulina Oliva, which is published in the March 2016 issue of the Economic Journal, has very important implications for policy. First, the results indicate that the important reductions in air pollution that environmental policy in Mexico City accomplished between 1997 and 2006 saved approximately 308 infant lives per year.

The findings also mean that the current levels of air pollution in the city are still dangerous – and that many lives could be saved from further reducing these levels.

Reliable measurements of the impact of exposure to air pollution on infant mortality are complicated since it is clearly not possible to run lab experiments exposing large populations of young children to pollutants. Instead, this study uses the next best alternative: a natural experiment created by meteorological events called ''temperature inversions''.

A temperature inversion occurs when a layer of hot air sits in between two layers of cold air. These events trap emissions near the ground, exposing people to high concentrations of air pollution.

To understand how these events lead to high emission concentrations, think of a hot air balloon. The same principle that leads a hot air balloon to rise above cooler air below it would normally lead emissions to escape the surface of the earth: warm gases are lighter than cool ones. But if on their way up the warm toxic emissions bump into a layer of hot air, this layer acts as a lid and prevents them from going further. Thus, gases accumulate near the surface of the earth.

This phenomenon happens from time to time and is relatively common in valleys like the one where Mexico City is located. Its relatively random occurrence provides a good source of experimental variation in the levels of air pollution concentration.

The new study uses data on the occurrence of this phenomenon on a daily basis along with air pollution data from 26 stations scattered around Mexico City and data on infant mortality.

Because temperature inversions are more common in Mexico City in the winter than in the summer months, the researchers ensure that the effects that they find stem from air pollution and not from other weather and seasonal causes that may coincide in timing with temperature inversions. They do so by comparing deaths across days with and without thermal inversions, but with very similar weather and within the same season.

Thus, the methodology is able to measure the effects that the additional tons of emissions stemming from a temperature inversion have on infant deaths. The results are large: when looking at the changes in carbon monoxide concentrations, the effect is 10 times larger than the effects found in other studies of carbon monoxide in California and New Jersey.

The researchers also examine whether the additional deaths include healthy infants or only infants that would have died from other causes shortly after. They find that the infants that die due to the additional tons of pollution in the air would have remained alive for at least one more month after exposure – and plausibly much longer.

It is worth highlighting that the results that this research measures correspond exclusively to short-run effects of air pollution exposure. Longer-run and cumulative impacts on health and mortality may also be present and are additional to the effects reported here.

''Does the Effect of Pollution on Infant Mortality Differ Between Developing and Developed Countries Evidence from Mexico City'' by Eva Arceo, Rema Hanna and Paulina Oliva is published in the March 2016 issue of the Economic Journal.

Eva Arceo is at the Centro de Investigación y Docencia Económicas (CIDE) in Mexico. Rema Hanna is at the Harvard Kennedy School. Paulina Oliva is at the University of California, Santa Barbara.