Research

Abstract: Governments and researchers often identify disaster-affected households us- ing self-reported exposure. This paper examines whether such reports re- flect physical exposure or broader household circumstances. I link Haiti’s post-2021 earthquake social registry to geocoded ground-shaking intensity and compare households within the same commune. Stronger shaking in- creased coping strategies, particularly severe ones. Yet households reporting earthquake damage used 0.8 more strategies than non-reporters, with shak- ing differences explaining only 3% of this gap, suggesting that self-reports conflate physical exposure with broader household distress. This conflation complicates impact measurement and targeting. Comparing hypothetical rules selecting 20% of households, self-reported catastrophe exposure iden- tifies only 20–23% of distressed households, barely outperforming random selection. A score based on registry characteristics identifies 26–32%, even when estimated using a small sample. These findings highlight the limita- tions of self-reported exposure and the potential of administrative data to improve disaster assistance targeting.

Abstract: The rigorous evidence on cash transfers and their accompanying measures (the“plus” components) is vast, but most of it comes from relatively stable settings. Testing how the evidence holds up in a more fragile setting is informative about how “settled” solutions translate when applied (or scaled up) in places where the evidence base is not usually generated. This paper evaluates Klere Chimen, the Haitian government’s safety net program, which paid about USD 36 a month and provided accompanying measures in Grand’Anse. The department is geographically isolated, recently hit by an earthquake, exposed to hurricanes, and, like much of current Haiti, linked to the capital by poor, gang- controlled roads. A single threshold on a continuous vulnerability score allows a regression discontinuity evaluation of the program, and the accompanying measures were randomly assigned to 26 of 52 sub-communes. Near the threshold, after about 20 monthly payments, eligibility raises mobile money use by 25 percentage points and food consumption scores by 9 percent (0.20 SD), close per dollar to the literature’s average effect on food security, while food spending rises by only about 34 cents per dollar. Among women in cash-eligible households, the accompanying measures raised savings-group membership by 12 percentage points and borrowing by 9, but may have offset much of the food consumption gain. We document no productive effects of the program, but suggestive evidence that children attend school more, women are more likely to report owning livestock, and households are less likely to have members migrate.

Abstract: Extreme heat threatens the labour supply and earnings of daily-wage workers in de- veloping countries, where physically demanding outdoor work is common. We examine these effects among workers employed under India’s National Rural Employment Guar- antee Scheme (NREGS), one of the largest public employment programs globally. We combine administrative data on attendance and assessed wages with a discrete choice experiment to elicit worker’s preferences over alternative heat adaptation measures. A one-degree increase in mean wet-bulb temperature over the preceding ten days raises absence by 3.0 percent of its mean; absence is 6.7 percent higher in the hottest than in the coolest quartile of accumulated heat. Effects on assessed wages are weaker, al- though the probability of receiving the scheme’s full daily wage declines with increase in temperature during the hottest months. In the choice experiment, workers require compensation equivalent to 2.8 percent of the daily wage to accept a one-hour later start but are willing to forgo 5.8 percent for a 10 percent reduction in the work target, preferring reduced work requirements to changes in work hours.

Abstract: This paper studies how precipitation conditions affect children’s education in ur- ban Ghana and traces the household mechanisms behind the effect. Combining four waves of the Ghana Socio-Economic Panel Survey with CHIRPS precipitation data and exploiting within-individual variation in the Standardized Precipitation Index, we find that a one-standard-deviation deterioration in precipitation condi- tions reduces children’s school attendance by roughly two and a half hours per week, about nine percent of instructional time. The loss is not recovered at home as homework time falls by a further 0.6 hours, close to thirty percent of its base- line. Enrollment, near-universal under Ghana’s fee-free basic education, and child health show no response. Point estimates indicate that households trim the non-fee components of educational spending, while school fees, zero for most children by policy, do not respond. The squeeze runs through two channels: drought lowers adult weekly earnings by about twelve percent, with point estimates nearly twice as large for men and no offsetting rise in employment or working hours, and it pushes households off free water sources onto paid alternatives, raising the unit price of water against highly inelastic demand. Chore hours rise in a small panel subsample, so children’s time absorbs part of the adjustment. Because we observe inputs into schooling rather than learning, our estimates speak to the resources devoted to education rather than to its output. Fee abolition alone does not insure schooling against climate shocks; protecting attendance requires drought-contingent support for the recurrent non-fee costs of participation.

Texts don ’t nudge: An adaptive trial to prevent the spread of COVID-19 in India – A reproducibility and replicability assessment of Bahety et al. (2021), with Florian Neubauer, and Julian Rose

Abstract: Bahety et al. (2021) evaluate an adaptive randomized controlled trial of SMS messages designed to promote COVID-19 preventive behavior in rural Bihar, India. The intervention finds no statistically significant effects on knowledge or self-reported social-distancing and handwashing behavior, which the authors interpret as evidence that SMS messages are ineffective in changing these outcomes. We computationally reproduce the results. In a multiverse analysis varying outcome definitions and control variables, all four main outcomes remain statistically insignificant. The original authors acknowledge that their statistically insignificant estimates cannot rule out potentially meaningful effects. We therefore complement the robustness analysis with equivalence tests to assess more systematically which effect sizes can be ruled out. When defining meaningful effects as effects exceeding 0.15 standard deviations of the control-group outcome, we can statistically rule them out for five of eight treatment–outcome combinations. For the remaining three combinations, the evidence is inconclusive: effects larger than this threshold cannot be ruled out. At more restrictive thresholds, the evidence is weaker. Overall, the original null findings are robust across alternative specifications, although the equivalence tests do not establish that all treatment effects are negligibly small.

Abstract: We study how rural households in semi-arid India reallocate labour over the months follow- ing two distinct weather shocks, rainfall deficits (drought) and extreme heat, using the high- frequency Village Dynamics in South Asia (VDSA) panel (2010–2014) merged with India Me- teorological Department gridded weather data, and the local-projections framework of Jordà (2005, 2023) adapted to a micro panel following Crofils et al. (2024). Two patterns emerge. Extreme heat is contractionary, it lowers total days worked and public-works (MGNREGS) participation and sharply curtails off-farm work during the heat-sensitive rabi (winter-crop) harvest, with the largest effects on women. Rainfall deficits instead induce reallocation: when the monsoon fails during the labour-intensive cropping season, local hired-farm work con- tracts and men shift toward out-migration, work outside the village, and public works three to five months later. Mobility is overwhelmingly male; women adjust in place through idle days and domestic work. On the demand side, on-plot labour-hours fall by roughly a fifth four months after a heat shock that strikes during the harvest stage, an effect absent during the grow- ing stage and borne mainly by hired labour, confirming that the contraction operates through harvest-stage labour demand. Timing of the response is governed by the agricultural calendar, identical shocks produce sharply different responses depending on whether they strike during the cropping season or the fallow.