How much do our neighbors really know?: The limits of community-based targeting (with Gumilang Sahadewo and Yudistira Hendra Permana; Job Market Paper)

Now split into 2 manuscripts: Main Paper and Better Together?

World Bank Development Impact Blog Post

Presentation

A classical motivation for using information provided by the local community to target social benefits in developing countries is that community members may have more current, dynamic welfare information about others than a centralized program implementer. However, there is little direct evidence supporting this claim; most relies on correlations between community-provided information and survey-collected welfare metrics. To understand the information community members have and use in targeting, we conduct lab-in-the-field experiments and community meeting exercises with 300 families in Purworejo, Central Java. Participants complete tasks where they individually rank other community members based on specific welfare benchmarks (consumption, neediness, and assets), as well as targeting tasks. We find that community-held welfare information is distinct from information captured using standard survey methods, and seems to reflect longer-term fixed attributes, rather than dynamic welfare information. Accordingly, community members use longer-term wealth information to predict dynamic welfare, and to target social benefits. Moreover, we find that community information about more dynamic measures does not outperform simple proxy means test scores in predicting more dynamic survey welfare metrics.  Finally, we find community members’ information sets are fairly concordant, and rankings constructed during community meetings do not seem to more closely reflect survey-collected welfare metrics. These findings suggest that community-based targeting methods may be useful in identifying long-term poverty, but less useful in identifying acute short-term distress. 

The Onion Value Chain in Senegal: Alternative Configurations and Differential Performances (with Alain de Janvry, Abdoulaye Cisse, Elisabeth Sadoulet, and Mame Mor Syll Anta)

Recent agricultural development literature has emphasized a more holistic “value chain” approach to food systems interventions. Given the apparent complexity of value chain structures in developing countries, much of this work studies specific bilateral links between value chain actors in distinct pathways (where we define a pathway as a sequence of actors among whom a good is exchanged from the producer to the final consumer). However, such approaches may not paint an accurate picture of what occurs in the value chain as a whole and may miss interactions between activity in different pathways. In this paper, we develop an approach to characterize the ecosystem of pathways that exist within a value chain, as well as characterize the relative importance of each pathway in terms of overall transaction volume. We apply our approach to the domestic onion value chain in Senegal, which yields useful insights about the value chain’s structure and functionality. First, we find that the longest, most complex pathways only comprise a small share of the overall volume transacted in the value chain. Second, we find actors’ choices to participate in various value chain pathways are adaptable to heterogeneity over space and time. However, smallholder producers may still be excluded from the least complex chains given their limited production volumes. Third, we find evidence that a quality-price premium is at least partially passed through to producers regardless of the pathway in which they participate, even though producers in more complex chains suffer from losses due to under-reporting of sale prices by the intermediaries who sell on their behalf.

Understanding Gender-Specific Constraints to Agricultural Technology: Evidence from Cassava Farming in Kenya (with Ethan Ligon and Muthoni Ng'ang'a)

Female subsistence farmers in developing countries often have lower adoption rates of agricultural technologies. These lower adoption rates may be due to lack of physical or informational access to new technologies, among other explanations. In this study, we consider these two classes of explanations of low technology use among females, and consider the relative impacts of interventions designed to combat each. We consider the technology of improved cassava in Murang'a County, Kenya, a more climate-resistant maize substitute. Using a randomized control trial with a 2x2 matrix treatment design, we test the effects on cassava adoption by female farmers of two interventions: delivering cassava seeds directly to female farmers at their homes (improved access), and hiring female ``lead farmers,'' to diffuse information about cassava seeds (improved information access), as well as explore complementarities between these interventions.

Interventions to Accelerate Varietal Turnover: Production vs. Consumption Oriented Approaches (with Gashaw Abate, Prakashan Chellattan Veettil, Beliyou Haile, Julius Juma, Berber Kramer, Catherine Ragasa, Bjorn van Campenhout, and others)

Smallholder producers throughout the developing world commonly grow old seed varieties, despite the availability of newer alternatives. One common explanation is that it can be risky for farmers to experiment with new varieties, as newer varieties are often more expensive, and farmers may be unsure of how such varieties will perform on their land. However, smallholder farmers are also often the main consumers of their production, and hence they might face additional consumption-related risks when choosing to grow a new variety (i.e. the risk that they dislike the taste/cooking quality of the new variety). In this project we compare interventions that address production-side risks (free seed trial packs) and consumption-side risks (free samples of crops produced) to adoption of new varieties. Specifically, we compare the effects of these interventions using a randomized control trial with a 2x2 matrix treatment design, where farmers either receive the seed trial packs, the sample crops, both or neither. We carry-out this design across five different countries (Ethiopia, India, Kenya, Nigeria, and Uganda) considering different crops in each setting.