Learning About a Warming World: Attention and Adaptation in Agriculture

Abstract

Global warming threatens the livelihoods of 600 million low-income agricultural workers. I study how farmers learn about the environment and the consequences for climate change adaptation. Rice farmers in Bangladesh must form beliefs about local soil salinity, a climate danger exacerbated by rising sea levels that can be mitigated by planting salinity-tolerant seeds. Comparing beliefs about salt levels to agronomic readings, I find that on average farmers hold correct environmental perceptions, but this masks substantial errors across individuals. I explain this pattern using a conceptual framework of belief formation featuring an identification problem: farmers must learn about multiple unobserved environmental threats from ambiguous signals. As a result, farmers endogenously process data in support of their priors, e.g., someone worried about high salinity will interpret low yield as a sign of too much salt. Climate change amplifies this process by systematically altering the environmental risks farmers consider most threatening. I test and confirm the framework’s predictions using a pair of natural experiments that capture two changes emblematic of global warming: salient shocks that capture attention (e.g., tidal flooding) and subtle shifts that go unnoticed (e.g., groundwater contamination through rising sea-levels). Despite equal effects on true salt levels, I find asymmetric impacts on beliefs as subtle groundwater intrusion causes no change in perceptions while saltwater floods spur significant overestimation. Analyses of rainfall and flooding perceptions exhibit similar patterns to those of soil salinity. In large-scale field experiments, I document major economic consequences of environmental beliefs: correcting misperceptions significantly alters farmers’ demand for salinity-tolerant seeds with substantial impacts on profits. I use this experimental variation to estimate and validate a structural model of seed choice that allows me to simulate counterfactual policies and underscores the influence of environmental beliefs.