Research

Communities worldwide frequently confront sudden, disruptive challenges that underscore the importance of resilience: the extent to which individuals cope with adversity. Governments, NGOs, and companies have implemented programs to aid the most vulnerable, yet uptake of such programs remains strikingly low. Through a randomized controlled trial with 1,048 women in rural India during the COVID-19 pandemic, we compare sales employment to an income-equivalent placebo and a no-employment control. Sales employment significantly increases uptake of formal relief programs, an effect especially pronounced for women from the most marginalized castes. The benefits prove durable: two years later, sales participants report significantly higher household income and well-being. Converging evidence points to the social connections forged through outward-facing sales activity as the most plausible driver. This investigation offers a novel perspective on the causal relationship between sales employment and resilience, highlighting the potential of sales roles to build connections that enhance personal and professional lives.

This study explores how zero-interest loans, in the form of inventory, can empower rural women entrepreneurs, selling socially beneficial products, economically and socially. The study also explores the relative efficacy of such loans versus general marketing support to make them better salespersons. The study used a  randomized controlled trial with four arms across three states in India (Rajasthan, Uttar Pradesh and Bihar): a group with only inventory loans, a group with only marketing support, a third group with both, and a final control group that operates under ‘business as usual’ conditions. While the group with both loans and marketing support is expected to perform the best, the relative efficacy of loans versus support is an open question. The results are expected to inform scalable models for rural entrepreneurship and financial inclusion.

Migrants own and operate a substantial share of the world's retail firms, yet their national identity may generate demand frictions that constrain firm performance. We provide the first estimates of consumer penalties against migrant-owned firms using a two-sided empirical strategy in Johannesburg, South Africa. On the demand-side, we implement an incentivized discrete-choice survey experiment (i.e., conjoint) in the field with $N = 1{,}973$ consumers. We randomize owner nationality alongside price, brand, and quality attributes, with an embedded information treatment to disentangle taste-based aversion from statistical discrimination. We complement this demand estimation with a supply-side census of retail firms in the same market ($N = 455$). We estimate a large demand penalty: consumers are willing to pay a 7.68 percent price premium to avoid a migrant-owned store, even conditional on disclosed store quality. Two mechanism tests support the taste-based interpretation of this residual penalty: it rises sharply with anti-migrant attitudes but is unrelated to consumers' quality beliefs and is not greater for perishable products as compared to non-perishables. On the supply side, migrant-owned stores charge 0.34 standard deviations lower prices on identical, branded products than matched local competitors. Together, the findings establish that identity-based consumer penalties distort equilibrium prices and compress profit margins for migrant entrepreneurs.

How can firms build relational contracts with smallholder suppliers at scale? Partnering with a leading global coffee buyer in Rwanda, we evaluate a low-cost relationship-building training delivered through the firm's existing extension staff, using randomized phone-call-versus-SMS invitations as an instrument for attendance. Attendance sharply improves the traceability of farmers' deliveries and strengthens their relationship with the firm. The central test is take-up of a tree-stumping contract that pays farmers conditional on delivering coffee cherries the following harvest, so that signing it means trusting the firm to honor a future promise. Attendance more than triples signing and raises verified compliance, while leaving take-up of an otherwise-identical, paid-on-the-spot contract unchanged. Exploring potential mechanisms, we find the training moved the trust margins our framework identifies — farmers' clarity about the terms of cooperation and, more modestly, the credibility of the firm's promises. Various robustness find no detectable effect on agricultural knowledge, practice adoption, or stable personality traits. Scalable relationship-building can thus create the trust that sustains relational contracts, and the costly investments they support, in agricultural supply chains.

Agroforestry turns farms into carbon sinks, shields coffee from extreme weather, and diversifies farmer income, yet adoption in poorer countries remains low. Payment-for-ecosystem-services (PES) schemes struggle with weak take-up when payments are delayed, poor incentives when payments are front-loaded, and the cost of monitoring whether trees survive. Interlinked transactions supported by long-term relationships often ameliorate such contracting problems in agricultural chains. Building on this insight, we partner with one of the world's largest coffee buyers, which sources directly from over 30,000 farmers in Rwanda, to co-design and evaluate an agroforestry PES scheme embedded within its existing sourcing relationships. In a randomized controlled trial with 3,711 registered farmers across five washing stations, participants receive free seedlings and are assigned to no subsidy, a standard upfront payment per tree planted, or a relational contract paying less upfront plus a survival bonus conditional on delivering part of the harvest to the buyer. We measure take-up, planting, and one-year tree survival, and test whether the relational contract also deepens the firm's sourcing relationship, turning sustainability from a cost into a commercial asset. The design offers a scalable template for embedding climate mitigation in agricultural value chains.

Automation is rapidly transforming manufacturing in developing countries. While it raises firm productivity, it has also reduced labor's share of the gains and now threatens the livelihoods of low-skilled workers. Yet the consequences of automation-induced mass layoffs, though increasingly common, remain largely understudied, particularly in low-income settings where formal insurance is scarce and labor markets are highly informal. This project examines how automation-related displacement unfolds in such contexts and how policy-relevant mitigation strategies can be designed. Working with a multinational exporter in Burundi that will lay off approximately 2,500 low-skilled female workers, we study the labor market, welfare, and behavioral consequences of mass layoffs, from reskilling and labor-market reintegration to potential reputational backlash for the automating firm. The project is structured around three research questions. First, how do workers form beliefs about automation risk, and how do these beliefs translate into labor supply decisions, job search behavior, and up-/re-skilling investments? Second, how do social networks and informal insurance mechanisms buffer or amplify the shock of mass lay-offs, and what social protection schemes are most effective? Third, can firm-led interventions, such as information signals, job-matching, or partial job retention, mitigate the negative welfare effects for displaced workers?