人力资本形成的即时中断：来自马拉维农村高频家庭和地理空间数据的证据（英）2025

11202 Nowcasting Disruptions to Human CapitalFormation Evidence from High-Frequency Householdand Geospatial Data in Rural MalawiPublic Disclosure Authorized Elizabeth J. TennantAleksandr MichudaJoanna B. UptonAndres ChamorroRyan EngstromMichael L. MannDavid NewhouseMichael WeberChristopher B. Barrett Human Capital Project &Development Data GroupSeptember 2025 Policy Research Working Paper11202 Abstract Exposure to extreme weather events and other adverseshocks has led to an increasing number of humanitariancrises in developing countries in recent years. These eventscause acute suffering and compromise future welfare byadversely impacting human capital formation amongvulnerable populations. Early and accurate detection ofad- verse shocks to food security, health, and schooling iscritical to facilitating timely and well-targeted humanitar-ian interventions to minimize these detrimental effects. Yetmonitoring data are rarely available with the frequency andspatial granularity needed. This paper uses high-frequency household survey data from the Rapid Feedback Monitor-ing System, collected in 2020–23 in southern Malawi, toexplore whether combining monthly data with publiclyavailable remote-sensing features improves the accuracyof machine learning extrapolations across time and space,thereby enhancing monitoring efforts. In the sample, ill-nesses and schooling disruptions are not reliably predicted.However, when both lagged outcome data and geospatialfeatures are available, intertemporal and spatiotemporalprediction of food insecurity indicators is promising. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about developmentissues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry thenames of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely thoseof the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank andits affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. NowcastingDisruptionstoHumanCapitalFormation:EvidencefromHigh-FrequencyHouseholdandGeospatialDatainRuralMalawi∗ ElizabethJ.Tennant†AleksandrMichuda‡JoannaB.Upton†AndresChamorro§RyanEngstrom¶MichaelL.Mann¶DavidNewhouse§MichaelWeber‖ChristopherB.Barrett∗∗ 1Introduction Rural households in low- and middle-income countries are exposed to a wide variety ofadverse shocks, including weather extremes, disease and pest outbreaks, price fluctuations,and job losses. These events cause acute suffering and often compromise future welfare bydisrupting human capital formation among vulnerable populations who lack the resourcesor social safety nets to cushion these shocks. This adversely affects long-run prospects forinclusive and equitable economic growth and development (Adair et al., 2013; Acemogluet al., 2014; Rossi, 2020; Rossi and Weber, 2024). Improved real-time monitoring of food security, health, and education indicators canassist in detecting when and where human capital formation is disrupted. When combinedwith information on current human capital stocks and the infrastructure that supportstheir development—such as food systems, health services, and education—this can helpto guide interventions that prevent unnecessary suffering and long-term harm.However,high-quality, up-to-date household survey data with broad geographic coverage are rare.Established household surveys, such as the Demographic and Health Surveys (DHS) andthe Living Standards Measurement Study (LSMS) surveys, provide a rich picture of healthand well-being, but are collected infrequently and published with significant lags.Thesedata may therefore overlook critical periods or be published too late for use in respondingto crises as they unfold. High-frequency household surveys are emerging as one way to more nimbly detect trendsin socioeconomic indicators. However, high-frequency surveys come with their own trade-offs. The costs associated with more frequent revisits constrains sampling and may requirechanges in the mode of data collection, such as shifts to phone-based surveys that may notaccurately reflect conditions for the poorest households due to selection or reporting biases(Abay et al., 2022; Abate et al., 2023; Dillon et al., 2025; Gourlay et al., 2021). Costs and concerns about respondent burden, particularly for panel (i.e., longitudinal) surveys, alsolimit the length and sample size of high-frequency instruments (Abay et al., 2022). Where costs, capacity, or access constrain the availability of survey-based data, oneway forward is data fusion: integrating complementary data streams to construct a morecomplete picture. A growing literature seeks to fill gaps in the spatial or temporal cover-age of hou