Strengthening bureaucrat incentives can curb crop burning and save lives in India and Pakistan
Reprinted by permission from VoxDev
Evidence from India and Pakistan shows that harnessing district officials' local pollution incentives reduces crop fires by up to 14.5% and deters burns by a further 13%, significantly lowering infant and child mortality.
Air pollution in South Asia is one of the largest public health crises on the planet: wintertime PM2.5 routinely soars to ten times the WHO safety threshold, claiming nearly two million lives each year and inflicting irreversible damage on child development, educational attainment, and economic productivity (Besley et al. 2022, Heft‑Neal et al. 2020, Khanna et al. 2025). Crop‐residue burning-farmers' low‑cost method for clearing fields after the rice harvest-drives 40-60% of those peak concentrations despite legal bans in India and Pakistan (Irfan 2023). These actions bear negligible direct costs to farmers despite severe respiratory, cardiovascular, and developmental harms on downwind communities. South Asian state administrators are thought to have little to no impact over crop burning, which is increasingly considered intractable (Jalali 2023, Srinivasan 2023). Our research (Dipoppa and Gulzar 2024) shows that tackling this externality does not necessarily require creating new agencies, but rather a smarter use of existing bureaucratic structures.
Figure 1: Air quality in Lahore, Pakistan

Note: The photograph on the left is from 12 April 2020, when the air quality index was 37. The photograph on the right is from 2 January 2020, when the air quality index was 496. Photograph credit: Dawar Hameed Butt.
Incentivising bureaucrats to deter crop burning
We use changes in wind direction as externally imposed shifts in bureaucrats' incentives to prevent or allow crop burning. When winds carry smoke into an official's own district, they face direct pressure-from voter complaints to performance evaluation. However, when smoke blows into other jurisdictions, there is less incentive for enforcement.
Drawing on ten years of monthly satellite fire data (2012-2022) and wind records for 5 km² grid cells in 207 districts across India and Pakistan, we estimate how fire counts change when a grid's smoke shifts from polluting a neighbour to its home district using a difference-in-differences framework. We find the following:
- When smoke hits home, there are 14.5% fewer fires. Our interpretation of this is that bureaucrats enforce the ban more aggressively when their own constituents are affected, but are more willing to allow crop burning when they do not internalise the pollution externality.
- Stronger effects at borders and in rice areas. Where capacity is higher or incentives sharper-along district or national boundaries and in rice‑dominated regions-the strategic fire reductions are even larger.
- Punishment creates deterrence. Bureaucratic action against farmers (for example, through fines) in treated areas lead to a further 13% drop in subsequent crop burning, suggesting that if bureaucrats can be incentivised to act, the benefits are likely to be greater than the direct effects of exerting control.
Our results demonstrate that existing bureaucratic structures-if properly leveraged-can substantially curb an entrenched environmental hazard without new institutions or subsidies.
Figure 2: Event‐study of crop fires around wind‑driven incentive switches

Note: Relative fire counts (with 95% CI) fall only when smoke enters the bureaucrat's home district.
Crop burning is extremely detrimental to infant health
To link enforcement to human health, we merge fires to geocoded Demographic and Health Survey birth records (n≈542,150) and use an atmospheric model, HYSPLIT, to predict how pollution particles from fires disperse across space. We estimate the total PM2.5 particles exposure from upwind fires each child in our sample experiences in utero and apply it as an instrument in an instrumental variables framework to measure the causal impact of greater air pollution exposure on infant and child mortality. Our estimates show that doubling expecting mothers' exposure to smokes leads to 21 additional child deaths in every 1,000 births.
We conduct a back-of-the-envelope calculation to link administrative action to health outcomes. Our estimates suggest that 1.8-2.7 deaths per 1,000 (equivalent to 4.4-6.6% of the average child mortality rate) could be prevented if bureaucrats were able to reduce crop burning to the levels observed in areas where they are directly affected by the pollution. Our estimates show that strengthening bureaucratic incentives does more than clear the air-it can directly save young lives on a large scale.
Figure 3: Additional infant and child mortality per 1,000 births from a 1‑log rise in in utero PM2.5

Note: Blue bars show point estimates; dark lines mark 95% CI. The effects reflect the increased incidence of child infant and child mortality per 1,000 births from in utero exposure to 1 log point more air pollution.
Policy implications for reducing air pollution
Our analysis shows that cleaner air need not wait for new agencies or costly programmes-rather it requires giving existing administrators a more direct stake in pollution outcomes. For example, district administrators could be held directly accountable for local pollution by integrating real‑time satellite fire counts and PM2.5 alerts into their performance reviews and promotion criteria.
We also find that penalties deter not just the farmers fined but others who might anticipate similar action against them, amplifying deterrence well beyond the farmers actually fined. A strategic campaign-combining swift, transparent and targeted fines with low‑cost subsidies or microcredit for residue‑management tools-could jointly push farmers toward cleaner alternatives without imposing an excessive burden on farmers.
By reframing crop burning as an accountability challenge rather than a technological dilemma, governments may be able to unlock larger gains in air quality and child health-leveraging existing structures to deliver lasting impact.