The (Non)Separability of Air Quality: Evidence from Millions of Households Across the United States

The costs and benefits of environmental policy depend crucially on the assumed microelasticities between market and non-market goods. In their absence, general equilibrium models have assumed environmental amenities are perfect substitutes with market goods, such as consumption and leisure, producing qualitatively dierent welfare assessments of environmental policy under even a narrow range of micro-elasticities. I estimate these elasticities using over 40 million observations from Census micro-data, together with weather and air quality measures at the county-level, between 2000-2014, finding that the elasticities between air quality and consumption, housing, and leisure are 7.14, .54, and .2, respectively. These estimates are identified from county-industry-specific deviations in air quality from the county averages after conditioning on shocks common to all counties within a state. Under simulated counterfactual distributions for 2010, these elasticities imply that the Clean Air Act Amendments had very large negative eects.