Research

Working Papers

• "Climate Change, Conflict, and the Risk of Economic Collapse in Africa" (forthcoming @ PNAS)

  with Marshall Burke, Edward Miguel, and Solomon Hsiang

  VoxDev

  Is human-caused climate change likely to trigger enduring economic decline in modern societies? We study whether changes in violent conflict and economic growth caused by warming could interact to degrade economic opportunities and trap African countries in poverty. We provide evidence of an emerging high-conflict and low-growth "poverty trap" equilibrium in Africa and describe theoretically how such an equilibrium could result from warming. We then combine historical data on temperature, growth, and conflict with projections from a large ensemble of global climate models to evaluate the risk that warming and subsequent conflict could push African economies into a regime of sustained negative economic growth. We find that the risk of such "economic collapse" is material in a moderate-to-high emissions scenario. We estimate that in Africa, a current "high" emissions trajectory (RCP 7.0) may increase the incidence of conflict 5.1 percentage points (95% CI 0.2-13.0) and increase the share of countries with net negative GDP growth this century by 12.2 percentage points (95% CI 2.0-27.5). We calculate that roughly 82% of this additional conflict risk and 14% of projected GDP losses are due to the interacting effects between these two outcomes, underscoring the importance of accounting for their indirect effects and feedbacks. Our findings suggest an unprecedented scale of emissions mitigation, economic policy innovation or institutional investment that would be required to contain the risk of catastrophic human impacts from climate change in many African countries.

• "Conservation Agriculture Diffusion Generates Large Human Health Benefits"

  with Andrea Gardeazábal-Monsalve and Bram Govaerts

  Despite being employed on more than 9% of global cropland, little is known about the environmental consequences of sustainable agricultural intensification techniques as practiced by farmers in their own fields. This paper studies the consequences of the rapid diffusion of conservation agriculture across the Mexican countryside on air quality and infant mortality in nearby urban localities over the period 2009-2022. We combine data collected by extension technicians on the timing and location of conservation agriculture adoption with daily wind direction variation to isolate quasi-random variation in localities' exposure to conservation agriculture and estimate causal effects. We find that at the mean level of exposure to conservation agriculture diffusion in 2022, fine particulate matter concentration falls by 13% and the monthly infant mortality rate falls by 10.9 deaths per 100,000 living. Effects on other air pollutants and seasonal heterogeneity suggest that these benefits are due to reductions in dust generated by land preparation activities. Aggregate benefits in terms of value of statistical life are an order of magnitude greater than the costs of extension provision, suggesting that conservation agriculture promotion has been a surprisingly cost-effective policy.

• "What Started China's Miracle? The View from Outer Space"

  with Oliver Kim

  We study one of the central reforms in China’s economic miracle, the Household Responsibility System (HRS), which decollectivized agriculture starting in 1978. The HRS is commonly seen as having significantly boosted agricultural productivity—but this conclusion rests on unreliable official data. We use historical satellite imagery to generate new measurements of grain yield, independent of official Chinese statistics. Using two separate empirical designs that exploit the staggered rollout of the HRS across provinces and counties, we find no causal evidence that areas that adopted the HRS sooner experienced faster grain yield growth. These results challenge our conventional understanding of decollectivization, land reform, and the origins of the Chinese miracle.

• "Downscaling Crop Yield Estimates Using Geospatial Foundation Embeddings"

  with Kangogo Kibarar and James E. Sayre

  Crop production statistics are critical to food security planning, yet they are costly to collect and are typically reported at large administrative units such as states, counties, or municipalities. Existing methods that use satellite imagery to produce high-resolution yield estimates rely on complex pipelines and/or disaggregated ground truth that may be inaccessible to practitioners and researchers. Our study develops an alternative approach that uses histograms and quantiles of Google AlphaEarth Foundations (AEF) embedding distributions to downscale yields using only publicly available aggregate data. We apply our model to downscale municipality (admin-2) maize yields to near-farm scales and validate its performance using disaggregated ground truth from the Mexican Agricultural Census, attaining an overall R2 of 0.61 and within-R2 of 0.23. We find that our best model, trained with AEF features, explains 78% of the within-municipality yield variation that can be predicted by a model trained on the ground truth labels. Approaches of the kind used in the literature have limited within-municipality prediction performance, with within-R2 below 0.10, less than half that of our best AEF specification. We highlight within-region R2 as a complementary metric for evaluating whether models recover fine-scale yield variation beyond what is already observable in regional-level official statistics. To aid in this evaluation, we propose a set of diagnostics to help others anticipate downscaling performance in the absence of disaggregated ground truth data.

Published

• "Irrigation Infrastructure and Satellite-Measured Land Cultivation Impacts: Evidence from the Senegal River Valley" Journal of Development Economics, 2026

  with Abdoulaye Cissé, Alain de Janvry, Marco Gonzalez-Navarro, Samba Mbaye, Elisabeth Sadoulet, and Mame Mor Anta Syll

  VoxDev

  Increasing the extent of irrigation in sub-Saharan Africa is a potential approach to closing yield gaps and improving resilience to climate change. Using over 3,000 satellite images spanning more than 30 years, we show that development of irrigation infrastructure in the Senegal River Valley has led to a large increase in average cultivation rates: a sixfold increase from 4 to 24 percentage points in the non-rainy season and a tripling from 10 to 30 percentage points cultivated in the rainy season. In spite of this substantial increase in cultivation rates, we find widespread heterogeneity across projects, with 1/4 of total irrigated area remaining unused as of 2019. We provide farmer survey evidence that limited access to water remains a major constraint on production and that removing this constraint requires action beyond individual farmers.

• "New Evidence on the Economics of Climate and Conflict" The Handbook on the Economics of Conflict

  with Marshall Burke, Solomon Hsiang, and Edward Miguel

  NBER Working Paper

  Press Coverage: Axios

  VoxDev/VoxEU

  This chapter surveys the quantitative research literature linking climate and conflict, a body of research that spans multiple academic disciplines and has roughly doubled in size in the last decade. It makes three main contributions. First, we carry out a meta-analysis - updating Hsiang et al (2013) and Burke et al (2015) with a far larger sample - and confirm that extreme climate is associated with elevated risk of inter-group conflict, intra-personal violence, and self-harm. The estimated average effects are smaller than the earlier estimates, although they remain meaningful in magnitude and highly statistically significant. Second, we present a methodological discussion and empirical illustration of how the use of data at different spatial and temporal scales can affect these results. Third, we discuss specific studies that present evidence on the mechanisms underlying this relationship including: (1) economic conditions, income, and agricultural productivity; (2) socio-demographic factors; (3) migration and transportation costs; (4) policy, politics and institutions; and (5) psychological and physiological factors. The chapter concludes with a discussion of policy implications and some open questions.

• "Will Wealth Weaken Weather Wars? AEA: Papers and Proceedings, 2024

  with Marshall Burke, Solomon Hsiang, and Edward Miguel

  This study estimates the moderating impact of economic development on climate-conflict linkages during 1989–2019 in Africa, the world region that in recent decades has experienced the most armed conflict. We build a spatially disaggregated dataset that merges multiple decades of georeferenced data on climate shocks and conflict events with both local- and national-level measures of economic development to help shed light on the relative importance of local opportunity costs versus state capacity. We find that higher national GDP per capita greatly dampens the conflict risk associated with higher temperatures, suggesting that enhanced state capacity is a key factor.

• "Survey of open science practices and attitudes in the social sciences" Nature Communications, 2023

  with Rebecca Littman, Garret Christensen, Elizabeth Levy Paluck, Nicholas Swanson, Zenan Wang, Edward Miguel, David Birke, and John-Henry Pezzuto

  Open science practices such as posting data or code and pre-registering analyses are increasingly prescribed and debated in the applied sciences, but the actual popularity and lifetime usage of these practices remain unknown. This study provides an assessment of attitudes toward, use of, and perceived norms regarding open science practices from a sample of authors published in top-10 (most-cited) journals and PhD students in top-20 ranked North American departments from four major social science disciplines: economics, political science, psychology, and sociology. We observe largely favorable private attitudes toward widespread lifetime usage (meaning that a researcher has used a particular practice at least once) of open science practices. As of 2020, nearly 90% of scholars had ever used at least one such practice. Support for posting data or code online is higher (88% overall support and nearly at the ceiling in some fields) than support for pre-registration (58% overall). With respect to norms, there is evidence that the scholars in our sample appear to underestimate the use of open science practices in their field. We also document that the reported lifetime prevalence of open science practices increased from 49% in 2010 to 87% a decade later.

• "Incorporating High-Frequency Weather Data into Consumption Expenditure Predictions" NeurIPS ML4D 2021

  with Anders Christensen and Simón Ramírez Amaya

  Recent efforts have been very successful in accurately mapping welfare in data-sparse regions of the world using satellite imagery and other non-traditional data sources. However, the literature to date has focused on predicting a particular class of welfare measures, asset indices, which are relatively insensitive to short-term fluctuations in well-being. We suggest that predicting more volatile welfare measures, such as consumption expenditure, substantially benefits from the incorporation of data sources with high temporal resolution. By incorporating daily weather data into training and prediction, we improve consumption prediction accuracy significantly compared to models that only utilize satellite imagery.

• "Marketing and Technology Adoption and Diffusion" Applied Economics: Perspectives and Policy, 2020

  with Amir Heiman and David Zilberman

  Applied economists have investigated individual adoption choices as wellas diffusion (aggregate adoption). The emphasis, however, has been on adopters’ behavior and risk associated with production and markets. Marketing also considersbroader aspects and marketers develop tools to address risk related to thefit of a product, its performance, and its reliability. This paper expands the economic literature onadoption by analyzing and assessing the implications of the choice of marketing tools, like money-back guarantees, demonstrations, and others, by marketers. The analysisis based on the threshold model of diffusion, which recognizes heterogeneity and dynamics. We provide evidence and examples from agriculture.

Work in Progress

• "A High-Resolution Look at Long-Run Development: Evidence from 1.3 Million Historical Photographs" with The Aerial History Project Team

  Press Coverage: E&E News

• "Conflict and Agricultural Land Abandonment in Africa"

• "Markets and Climate Change Adaptation: Evidence from Minnesota and Wisconsin Farmland Transactions"