RESEARCH

In the literature, there is limited evidence on how cash reallocations within beneficiary groups respond to unanticipated shocks during the project implementation phase. In practice, reprogramming tends to be rigid due to logistical constraints, despite the need for flexibility in fragile environments. This paper employs a machine learning approach, specifically double selection for covariate selection, to evaluate whether cash reallocations targeting pregnant and lactating women in Myanmar were effective amid COVID-19 and armed conflict shocks. Using a Difference-in-Differences framework, the results suggest that the treatment was only effective when subgroups were analyzed. In particular, incorporating distance to conflict locations reveals an increase of between 4 to 5 points in food consumption score among households located farther from conflict zones, while no gains are observed among those residing within 4 km of conflict zones. Additionally, households that did not report experiencing any shock saw an increase in per capita food expenditure between 6 to 7 US dollars. Beneficiary households also reported reduced reliance on coping strategies. Notably, these effects are statistically significant only when machine learning methods are applied, and remain insignificant under conventional estimation techniques. I emphasize two key findings: (i) the need for adaptive social protection mechanisms in fragile settings and caution against relying sorely on subjective shock reports for vulnerability assessment, as is common in most humanitarian aid projects; (ii) the importance of machine learning techniques to achieve more precise estimates by controlling for high-dimensional confounders, enhancing the reliability of impact evaluations. 

We propose the Labor Exclusion Theory as a novel framework for understanding how structural or policy-induced exclusion of specific groups from formal labor markets generates inefficiencies in overall market outcomes. We test its short-term predictions in the context of Afghanistan’s female labor ban, focusing on the informal sector. 

Abstract available soon >> 

To aid or not to aid: The tug of war between efficient policy implementation and constrained resources.

With Nilima Pisharody and Lena Uzelac.

On migration, rural poverty, and agricultural performance: New evidence from machine learning methods.

(Recruiting a research assistant)

Uganda is home to some 1.5 million refugees. Its self-reliance policy welcomes refugees with freedom of movement and work; a piece of land for building a house and home-gardening; and social protection (food and cash transfers). While a progressive refugee policy, the onset of COVID-19 may complicate its effectiveness. This paper investigates the role of food and cash transfers in meeting negative consequences of the pandemic for both host and refugee communities. We do so by using a unique panel dataset covering the period from 2019 to 2021. Our evidence suggests that although transfers were widely used in refugee settlements prior to COVID-19, their use did not expand within the refugee nor host communities at the onset of the pandemic and, in fact, the transfer amounts declined. The evidence does not suggest that either form of transfer has a strong effect on food security or resilience although there is some evidence that cash transfers facilitate resilience. However, food transfers are associated with maintaining or increasing food security while cash transfers are more closely associated with resilience. Furthermore, food transfer seem to outperform cash transfers in responding to COVID-19 particularly when income loss due to COVID-19 is reported.

The recent COVID-19 global pandemic has revealed that despite numerous development efforts, there are still inefficiencies in maintaining the living standards of people when shocks and stressors occur. While addressing issues arising from the pandemic is dramatically urgent, this should not come at the cost of averting resources and efforts from sustainable and equal growth and prosperity goals. The importance of resilience for the humanitarian and development nexus, has probed United Nations agencies, international organizations, donors, and governments to investigate key facts and determinants of this capacity. After approximately 15 years of empirical evidence, few research questions remain unexplored and unanswered. Are there few and consistently relevant elements that determine resilience capacity? What shocks are most dramatically reducing resilience? What coping strategies are most frequently adopted in the presence of shocks? This paper attempts to respond to these questions by pooling together a unique database of 35 countries. This study combines the most recent FAO-RIMA (Resilience Index Measurement and Analysis) datasets with a large set of data from the Multiple Indicator Cluster Survey (MICS) produced by the United Nations International Children's Emergency Fund (UNICEF). The analysis covers the period between 2014 and 2020 by investigating 50,622 households. The size of the sample provides our findings with great statistical power, therefore adding external validity. 

The CGIAR Digital Innovation Initiative (2022–2024) is a transformative effort to leverage digital technologies in addressing critical challenges in global agrifood systems. By emphasizing inclusivity, sustainability, and scalability, the initiative has made significant strides across five key outcomes: strengthened ecosystems, improved digital skills, equitable resource allocation, gender-responsive services, and robust information systems. This report synthesizes evidence from case studies, documentation reviews, and stakeholder interviews, showcasing innovations such as Digital Twins, AI-powered advisory platforms, and gender-inclusive digital services. Findings demonstrate substantial impacts, such as empowering women and youth through gender-inclusive digital services, improving digital literacy, and enabling equitable resource management in climate- vulnerable regions through advanced tools. However, embedding digital innovations effectively and efficiently in land, water, and food systems remains a complex and context-specific challenge. The results highlight the need for ongoing research, testing, and validation to address the particularities of agri-food systems, such as data integration, socio-cultural barriers, and stakeholder diversity. Recommendations emphasize building on existing successes, expanding digital ecosystems, and fostering partnerships to secure long-term impact and sustainability.

Resilience measurement can now be viewed as an established body of research with 15 years of empirical evidence. Across this body of work, measurement studies have typically sought to identify key elements that render some households more resilient than others. There is now ample literature that includes robust and solid methods (Cissé and Barrett 2018; d’Errico, Romano, and Pietrelli 2018; Knippenberg, Jensen, and Constas 2019; Smith and Frankenberger 2018), reviews of methodologies (Barrett et al. 2021), solid evidence on the impact of resilience-enhancing interventions (d’Errico et al. 2020), and evidence on the role of macro and covariate shocks (such as conflict) on resilience capacity (Brück, d’Errico, and Pietrelli 2019).