Research

What are the causal effects of new social relationships, and do returns to link formation vary within versus across social class? I conduct a field experiment among women in rural Malawi who relocate to their husband’s home village, thereby disrupting their preexisting social ties. I randomly assign low socioeconomic status (SES) with the opportunity to send meal invitations to neighboring women willing to receive them. I experimentally vary whether the names appearing on the list are all low-SES, all high-SES, or both. I cross-randomize the invitation-sending arms with an in-kind meal subsidy, which weakens financial constraints to meal-sharing. Eighty-one percent of women send invitations across all treatment arms, while the meal subsidy increases invitations to high-SES women only among participants who can send invitations to both. I show that there are substantial but differential returns to cross- and same-SES linkages: one year later, participants with the opportunity to invite higher-SES guests are 20% more likely to earn income from self-employment and report a 0.21 SD improvement in food security, while women with the opportunity to invite guests within-SES experience a 0.23 SD (30%) reduction in depressive symptoms. This highlights a trade-off in network formation, where social ties that offer economic advancement may come at the expense of relationships that improve psychological well-being.

How does a person’s own learning experience affect their ability to learn from others? We conduct a field experiment on chlorination in Pakistan, where randomized learning-arm households use a tool to track their children’s diarrhea before and after chlorine distribution. Learning-arm households with learning-arm neighbors chlorinate significantly more one year after the withdrawal of the tool, with children’s health improving by 0.08 SD relative to all other households receiving chlorine. Neither learning households without learning-arm neighbors, nor non-learning households with learning-arm neighbors, exhibit sustained behavior change, results which have significant implications for intervention and evaluation design.

Female labor force participation is constrained by supply-side frictions (lack of qualified female applicants) and demand-side frictions (biases in hiring decisions). How do supply- and demand-side distortions interact to affect gender gaps in employment? We partner with a firm to conduct two experiments: The first experiment addresses supply-side behavioral gender gaps during a recruitment drive through female-directed advertising, while the second experiment is a resume audit study among evaluators to isolate biases in hiring. Despite no treatment effects on the objective skills of male or female applicants, the treatment has the perverse effect of leading to a reduction in female hiring. In the resume audit study, evaluators place greater weight on noisy signals of soft skills for women than men when shortlisting candidates, while, in the recruitment experiment, the treatment leads low-ability women to inflate their signals of soft-skills. Evaluators subsequently shortlist lower-ability women in treated areas, who crowd out higher-ability female candidates from the shortlisted pool. Our estimates suggest that mitigating either supply-side distortions in application signals or evaluator bias in signal weighting would close the gender gap in the ability of shortlisted candidates by 51-55%, with only modest additional gains to mitigating both.

We test a novel explanation for disparities in agricultural output between men and women in Malawi: child-bearing as a unique shock to women's labor supply that inhibits efficient land utilization. We implement a randomized controlled trial with female farmers who are pregnant or have a child under the age of two years old. We test the impact of providing five paid days of skilled agricultural labor during the farming season. For women who are enrolled in the trial during the first two trimesters of pregnancy, the program does not substitute for household labor, and, if anything, crowds in hired labor. These treated women experience a 0.26 standard deviation increase in harvest yields and a 0.39 standard deviation increase in maternal food security. For women enrolled in the trial during the third trimester of pregnancy, the treatment leads farmers to substitute away from hired labor and engage more in non-agricultural income-generating activities, with no effects on yields or food security. The program has no impact among women who are non-pregnant. We interpret these results through a model where the physiological labor supply shock of pregnancy interacts non-linearly with market failures and intra-household norms to create gendered shocks to farm productivity and consumption.

Why are people willing to pay for things that are symbolic? We analyze reactions to Indigenous renamings of U.S. geographic landmarks to argue that people lose utility through the way that symbolic change is enacted, even when they are indifferent to the symbols themselves. People living in zip-codes that experience geographic landmark name changes initiated by a Democratic administration are less likely to donate to Democratic political candidates after name removals are announced, but not when Indigenous replacement names themselves are announced. In a 2024 survey experiment, all respondents express much more conservative preferences about U.S. policy with Tribal Nations when we frame the 2015 (Democratic administration) renaming of Mt. McKinley to Mt. Denali as a process enacted by and benefiting groups of people ("active voice" treatment), rather than a passive occurrence. Emphasizing the Indigenous origins of the name "Denali" does not generate the same response. In a 2025 survey experiment, White respondents behave similarly, reporting weakly more conservative policy preferences when we frame the 2025 (Republican administration) renaming of Mt. Denali to Mt. McKinley as an active rather than passive process, while non-White respondents express much more liberal policy preferences in 2025 than 2024, flipping the sign of the active voice treatment effect. These results suggest that how people interpret the process of symbolic change depends on the interaction between personal identity and the identity represented by the symbolic change.

This paper investigates the role of peer influence in health behavior by studying antiretroviral therapy (ART) adherence among adolescents living with HIV in Malawi. With high-frequency electronic medical records of almost 45,000 adolescents, across twelve years, 85 clinics, and 23 districts in Malawi, we track high-stakes health decision-making for as long as adolescents receive care. Using a staggered differences-in-differences analysis of the effects of Teen Club entry at the clinic, we find that Teen Club entry increases on-time visit attendance by 33%. We validate this finding with a regression discontinuity design using age cut-offs in Teen Club eligibility and age at the time of initiation, and find that barely age-eligible patients are 65% less likely to ever be lost to care. Being barely age-eligible decreases the probability of ever contracting tuberculosis by 10%, and decreases five-year mortality by 140%. Finally, we utilize two instrumental variables specifications to measure how attendance at Teen Club is impacted directly by peer attendance in their prior Teen Club appointment, as instrumented by the peer's attendance record prior to the patient's initiation at Teen Club, or as instrumented by rainfall on the peer's appointment date.

Does exogenously generating social inclusion for one person have externalities for others' social inclusion? I leverage cluster-randomized treatment saturation of an experiment that creates exogenous opportunities for social linking, which effectively creates variation in the spatial proximity of Control participants to neighbors' network activity. The treatment design also randomly varies the socio-economic status of the exogenous interactions among low-income treated participants. This design allows me to answer fundamental questions about social network formation that we have limited causal evidence for: Does proximity to social network activity crowd in or crowd out one’s own social ties? Do these effects depend on socio-economic status of the agent who is proximate to others' social network activity, or the socio-economic status of the socially-active nodes to whom she is proximate?

We study how private health behavior responds to new public water infrastructure. During an 18-month randomized controlled trial (RCT) in peri-urban Karachi, we distribute chlorine tablets to households and collect monthly, objective measures of chlorine use and water source choices. Concurrently, local NGOs independently implement water infrastructure projects in a staggered fashion. This natural variation in proximity to new standpipes and boreholes, combined with randomized access to water treatment, allows us to estimate the dynamic effects of infrastructure exposure on private behavior and health outcomes.