This paper estimates the aggregate productivity gains from abatement of pollution from agricultural fires, accounting for both health-related labor productivity improvements and better spatial allocation of workers. If reducing pollution attracts marginal workers to more productive places, aggregate gains can come from improved spatial allocation of labor. To understand the relative importance of these mechanisms, I construct a spatial equilibrium model with migration frictions and pollution dispersal and apply it to India. Leveraging the labor supply equation predicted by the quantitative model, I estimate the income and pollution elasticities of migration using a Bartik instrument from international trade and exogenous variation in wind and fire patterns. I find that policy-driven abatement of agricultural fires in northwestern India during 2010 would have increased national income by 0.4%, with reallocation accounting for a modest but significant 18% of these gains. Relaxing high migration frictions increases the importance of reallocation. Relatively higher pollution reduction from fires near dense cities increases importance of reallocation by accentuating the agglomeration benefits of cities.

Industrial water pollution is high in many developing countries but often receives less attention than air and domestic water pollution. We estimate the costs of industrial water pollution to agriculture in India, focusing on 48 industrial sites identified by the central government as “severely polluted.” We exploit the spatial discontinuity in pollution concentrations that these sites generate along a river. First, we show that these sites do coincide with a large, sudden rise in pollutant concentrations in the nearest river. Then, we find that a remote sensing measure of crop yields is no lower in villages immediately downstream of polluting sites, relative to villages upstream of the same site in the same year. Downstream farmers switch irrigation sources from rivers and canals to wells in some specifications, suggesting costly input substitution may avert pollution damages. Damages to agriculture may not represent a major cost of water pollution, though many other social costs are not yet quantified.

Alarming rates of groundwater aquifer depletion in North India are linked to water-intensive rice cultivation based on cheap electricity for water pumps. In this second-best setting where optimal marginal pricing of groundwater is not possible, the northwestern states of Punjab and Haryana with the highest groundwater depletion rates instead instituted laws in 2009 intended to foster reliance on rain-fed irrigation by mandating a delay in rice crop transplantation to coincide with monsoon arrival. At the same time, rice crop residue burning in these two states contributes to high particulate matter levels over North India. In this paper, I use satellite data on monthly fires and a difference-in-differences framework to document that the groundwater laws shifted more than half of all agricultural fires into early winter, when meteorological conditions favor longer suspension of particulate matter over North India. I then quantify the consequences of this increased air pollution on Indian GDP by estimating two further elasticities. First, I develop a novel instrument for PM2.5 that summarizes the exposure of a given location to all upwind fires, showing that 10% higher district exposure to November fires increases annual PM2.5 by 0.3%, and that 4% of within-district annual variation in PM2.5 can be explained by exposure to November fires. Second, I estimate the effect of higher PM2.5 levels on GDP with data on Indian districts between 2007-2013 using district and year fixed effects combined with a first differences approach that is more efficient for non-stationary data, and with the fire exposure instrument to tackle residual reverse causality. With this approach, I find estimates that a 10% increase in PM2.5 reduces GDP by 1.8%, with a 95% interval of [-0.4%, -3.17%]. With these two elasticities and the structure of the instrument, I estimate that the groundwater laws decrease yearly Indian GDP by 0.125% due to the increase in November fire-driven air pollution.

Workfare programs such as India’s National Rural Employment Guarantee Scheme (NREGS) are an attractive way to better target consumption smoothing in the face of increasingly extreme weather events driven by climate change. Using the rollout of NREGS across districts in India with quasi-exogenous variation in yearly weather, I document increased volatility of crop yields after implementation of NREGS, with additional yield losses of 8% during a bad rainfall year post-NREGS. In order to test whether these results can be explained by the choice of higher-yielding but more volatile crops but due to the insurance properties of NREGS, I construct novel agricultural risk indices using pre-NREGS moments of district crop revenue distribution. Higher risk as measured by these indices is associated with higher crop yields in good rainfall years, but lower yields after negative rainfall shocks, therefore capturing meaningful features of aggregate risk in crop choice. Using the rollout strategy, I find little evidence that the increased sensitivity can be explained by these measures of risk in crop choice. On the other hand, I find that NREGS strongly dampens pro-cyclical wage response to low rainfall shocks, potentially exacerbating the yield effects of productivity shocks by increasing labor costs. Finally, higher provision of NREGS after a negative rainfall shock worsens yield losses if a negative rainfall shock is also realized next year, but improves yields if a positive shock is realized instead. Policymakers considering such programs should pay close attention to the negative and positive complementarities between social protection and agricultural productivity that these results suggest.