Nowcasting Disruptions to Human Capital Formation : Evidence from High-Frequency Household and Geospatial Data in Rural Malawi

Exposure to extreme weather events and other adverse shocks has led to an increasing number of humanitarian crises in developing countries in recent years. These events cause acute suffering and compromise future welfare by adversely impacting human capital formation among vulnerable populations. Early and accurate detection of ad- verse shocks to food security, health, and schooling is critical to facilitating timely and well-targeted humanitarian interventions to minimize these detrimental effects. Yet monitoring data are rarely available with the frequency and spatial granularity needed. This paper uses high-frequency household survey data from the Rapid Feedback Monitoring System, collected in 2020–23 in southern Malawi, to explore whether combining monthly data with publicly available remote-sensing features improves the accuracy of machine learning extrapolations across time and space, thereby enhancing monitoring efforts. In the sample, illnesses and schooling disruptions are not reliably predicted. However, when both lagged outcome data and geospatial features are available, intertemporal and spatiotemporal prediction of food insecurity indicators is promising.