Ethanol-Blended Gasoline Policy and Ozone Pollution in Sao Paulo

We examine four discontinuities in the ethanol content in blended gasoline fuel, mandated by Brazil's central government over the period 2010-13, to test the joint hypotheses that (1) atmospheric ozone production in the Sao Paulo metropolitan area is limited by the volume and reactivity of volatile organic compounds (VOC-limited) and (2) increased ethanol use in the gasoline-ethanol vehicle fleet leads to higher ozone concentrations in urban Sao Paulo's ambient air. We adopt a regression discontinuity design (RDD) and flexibly test each discontinuity separately. Our finding that ozone levels actually increased with ethanol penetration on each of the four occasions is consistent with a recent empirical study that used different identifying variation and contrasts with a modeling study of Sao Paulo's atmosphere that predicted significant ozone abatement from hypothetical ethanol use. We find no significant relationship between ethanol versus gasoline use and PM2.5 levels. Current tailpipe emissions standards prescribe the exclusion of the mass of unburned ethanol that is emitted--our results suggest that this standard be reviewed. Following decades of VOC emissions control, urban areas in the United States and elsewhere that are currently VOC-limited may see ozone levels rise if and when they adopt mid-level ethanol gasoline blends, whether to meet the Renewable Fuel Standard or Intended Nationally Determined Contributions to abating fossil fuel emissions agreed upon at COP21 in Paris.