Research
Natural disasters · Causal inference · Development economics
Job Market Paper
Proximity to the Plague: Measuring Locust Exposure and Child Health
Abstract
I investigate early childhood health following the Kenyan 2020–2021 desert locust invasion by linking georeferenced locust observations to a nationally representative health survey. To account for differences in harvest times across the country, I build a treatment that conditions on the insect’s biological capacity to damage vegetation, the growth stage of the local crop, and a subsequent decline in vegetation. Children exposed in utero to locust crop damage within 30 km are 25 percent more likely to be wasted than unexposed children. Among crop-growing households, the harm is concentrated in farming zones rather than those that also rely on herding or fishing, and among children born before the lost harvest. This pattern suggests that households cut consumption in anticipation of lost food. Relief may therefore be most effective if it reaches farming households as soon as crop damage is recorded, through fast-paying instruments such as cash transfers or expanded safety nets, rather than after the harvest fails.Working Papers
Wildfire smoke, air quality, and the limits of short-run labor-market impacts in Canadian cities With Christopher S. Cotton | Submitted: Proceedings of the National Academy of Sciences
Abstract
Wildfire-smoke-driven increases in fine particulate matter raise hours worked and employment in Canada’s nine largest metropolitan areas. This response is consistent with productivity impairment requiring longer hours to meet output targets and with degraded outdoor conditions reducing the value of non-work time. Both mechanisms imply welfare losses that standard labour-market statistics record as gains, suggesting output-based damage assessments understate the costs of smoke in urban economies.
Drought-Proof Generosity: Index Insurance and Informal Livestock Lending in Northern Kenya Last updated: April, 2026
Abstract
I investigate whether formal insurance contracts crowd out informal risk-sharing arrangements that underpin livelihoods in developing economies. Using panel data from pastoralist households in Marsabit, northern Kenya, I instrument insurance uptake with randomly distributed price discount coupons and interact coverage with sub-location drought severity. Drought erodes informal support for uninsured households. A one-standard-deviation worsening of drought lowers the probability that an uninsured household receives a livestock gift by 11 percentage points and a cash or in-kind transfer by 12 points, and lowers the probability that it gives a transfer by 11 points. Livestock borrowing does not fall. Two-stage least squares estimates suggest that the same worsening of drought raises livestock borrowing by roughly 40 percentage points more for insured than uninsured households. In the reduced form, coupon recipients also give and receive more transfers as drought deepens. I find no evidence that insurance crowds out informal risk sharing, and suggestive evidence that it helps keep informal networks active during a drought.
Work in Progress
Born into Drought: Early-Life Exposure and Child Health in AngolaAbstract
Abstract
Drought is widely expected to harm early childhood health, yet credible causal estimates remain scarce. Observational studies typically link child outcomes to moisture indices measured at survey-cluster locations, which are displaced for confidentiality and may poorly capture the water deficit each household actually experiences. Such measurement error, together with the endogeneity of where and when households are exposed, complicates interpretation of existing estimates. This paper explores a strategy for identifying the causal effect of early-life drought exposure on child stunting in Angola, using pooled DHS surveys from 2015-16 and 2023-24 that span the 2018-2021 southern African drought. I use low-frequency oscillations in atmospheric circulation, specifically the Southern Annular Mode, as a source of plausibly exogenous variation in moisture conditions, and I evaluate how alternative instrument definitions and exposure windows (in utero, first year, second year, and cumulative) affect the strength of the first stage. The project is a work in progress: I document the relevance and limitations of this approach, discuss threats to the exclusion restriction, and assess whether circulation-driven variation can improve on standard observational measures of drought exposure. The goal is to inform the targeting and timing of nutrition and water, sanitation, and hygiene interventions in drought-prone regions of southern Africa.What Replaces the Sun? Wildfire Smoke, Solar Losses, and Power Sector Emissions
Economic Shocks: Power Outages and Economic Activity. With Zachary Robb