Working papers (manuscripts available upon request)
Water contamination by Animal Feeding Operations: Evidence from Iowa
and North Carolina
The majority of U.S. animal farming takes place in industrial Animal Feeding Operations (AFOs), where large quantities of manure are produced, stored, then typically spread onto fields. The concentration of animal waste is thought to threaten environmental and public health, as airborne and waterborne transmission puts communities at risk of environmental exposure to its contaminants. Water pollution from swine operations is of particular concern yet there are no robust causal estimates of this externality to date. I gather panel data of all permitted AFOs in the two main hog producing states — Iowa (2002-2017) and North Carolina (1997-2020) — and estimate the effects of animal production on downstream surface water quality readings.
My analysis espouses the spatial structure of the underlying process by constructing station-specific drainage areas, and I use two identification strategies appropriate for the states’ regulatory settings and data resolution: the first leverages the spatiotemporal variation in animal production in Iowa, the second the randomness of intense precipitation events on North Carolina farms.
I find deleterious effects across water quality indicators and across types of operations, including from facilities below the size threshold at which the industry is currently regulated. A daily rainfall extreme at an AFO increases downstream levels of fecal coliforms by 3.5%, and nutrients by 0.5-0.7%, relative to mean sample levels, while an additional swine operation in the average station drainage area increases nutrient concentrations by 3.8-10.7% and decreases dissolved oxygen by 1.1%, relative to sample mean levels. These effects are higher than previous findings focused on large dairies, and are higher for swine AFOs relative to all AFOs.
Grounding Animal Farming
Animal farming is at a crossroads in industrialized economies, where three frameworks for potential system futures dominate the debate: sustainable intensification, agroecology, and abolition. This paper takes a systems perspective that sheds light on some of the limitations of each approach, and brings together insights from various disciplines to propose a unifying framework.
The argument first defines the boundaries and core components of the system of animal farming, and then identifies how each of the three frameworks addresses the interconnections between these elements. I bring forward key insights from research in agronomy, sociology and philosophy, which directly address blind spots of each framework, in particular the conditions for sustainable human-animal interactions. First, the sociological paradigm which identifies a work relationship between humans and animals as an essential part of animal farming. This lens reveals the multiple rationalities of animal husbandry and thereby also the limits of existing structures. Second, the extension of the notion of carrying capacity–typically used to qualify the amount of pressure that land can bear—-to the moral suffering that workers performing slaughter can endure.
To be sustainable, a framework for animal farming must address the interconnections of the three core components of this socio-ecological system: humans and their institutional structures, farm animals, and land. I articulate one which does: a grounded animal farming system. It recognizes the particular social relationships and the multiple carrying capacities at play as features that distinguish animal farming from other sectors of the economy. I show its parallels with proposed paradigms on how to pursue a larger socio-ecological transition.
Cow milk response to humid heatw. Eyal Frank, Ram Fishman, and Ayal Kimhi
Quantifying the effect of weather on economic outcomes is more important than ever given climate change projections. The methodological strengths of the climate-economy literature have enabled to precisely estimate response functions across sectors of the economy, but remain to be applied to animal agriculture. Humid heat stress is suggested to be one of the main limiting factors of milk production, however the external validity of existing estimates is limited by underlying linearity assumptions, low data coverage or resolution, and non-inclusion of adaptation potential. We use daily cow-level panel data on milk production, and survey data on adopted cooling technologies, to estimate the effects of humid heat on milk yield, in a setting conducive to large external validity. We find a gradually steeper decline with increasing levels of wet-bulb temperatures, reaching a 9.6% decrease for a day whose average is above 26°C, relative to the 10-12°C range. Such decreases embed both the contemporaneous effect of heat, and the sum of the delayed effects of previous serially-correlated hot days, that are still impactful over 10 days after exposure. We further find that marginal adjustments to the farming process—simple cooling technologies—are associated with reductions in the effect of heat of less than half. We observe a lower sensitivity to heat in less productive cows, suggesting a productivity-resilience trade-off. Under a new climate regime with more frequent and longer humid hot spells, the resilience of a system may become an increasing priority. Our results suggest that more structural changes may be required to reduce the vulnerability of animal agriculture substantially, such as reducing other stressors affecting cows in the dairy industry to reduce the compound effect on cow sensitivity.
A multilevel Bayesian framework to analyze climate-fueled migration and conflictw. Upmanu Lall, Paulina Concha Larrauri, and Andrew Gelman
Do climate conditions and extreme events fuel conflict and migration? This question, of growing interest given increasingly dire climate change projections, is commonly addressed in causal studies that leverage natural experiments by using a multivariate linear regression model with fixed effects. We show that in the climate-migration-conflict nexus, the features of the data generating process and the implicit prediction motivation make for a substantial departure from the assumptions of the typical linear reduced form model, which challenges the reliability of inferences. We propose a unifying hierarchical Bayesian framework for inferences from the same natural experiments, and describe its benefits for internal and external validity and for analyzing the heterogeneity in response to climate. We illustrate the misleading results that can ensue from the typical approach and the advantages of the hierarchical Bayesian framework by using a conflict dataset representative of the literature.
Work in progress
The Impacts of Right-To-Farm Laws on Agricultural Productivityw. Anthony D’Agostino and Eyal Frank