Climate change is raising groundwater salinity across coastal regions, threatening smallholder livelihoods. In Bangladesh, a common adaptation is to convert rice paddies into saltwater shrimp ponds — a technology that thrives where rice cannot. We ask whether this private adaptation generates negative externalities on neighboring rice farmers, and how this affects the adoption of the technology. Combining a novel household survey with a satellite panel covering  80,000 plots over nine years, we document a sharp salinity gradient at the shrimp–rice boundary: bordering rice plots show significantly higher salinity and lower yields than otherwise comparable plots. Event-study estimates show that a neighbor’s conversion from rice to shrimp raises own-conversion likelihood and depresses vegetation indices on remaining agricultural plots, suggesting that climate resilience for some can erode it for others.

International Growth Centre (IGC), STICERD, Hub for Equal Representation

Global warming will cause adverse changes to the climate faced by the world’s poorest. Individuals can mitigate the impacts of these shocks by adopting new technologies, working in less-exposed occupations, or migrating to safer areas. However, it is unclear whether exposed populations are sufficiently informed about future shocks to undertake these investments. We use a survey experiment (N=5,992) among a large sample of rural Bangladeshis to show that concerns about climate change are widespread; that individuals plan to invest in new technologies in-situ (rather than migration and occupation change); and that a large flood is associated with stronger demand for in-situ adaptation policies. Providing information about future shocks does not affect adaptation intentions or policy demand. Overall, the rural poor are highly concerned about climate change and would like vigorous investments in in-situ adaptation. Alternative policies aiming to support migration or occupation need to raise beneficiary interest to be effective.

International Growth Centre (IGC), JPAL-KCAI

Do the poor have high returns from trainable human capital? Can they afford to pay for it? And are they differentially targeted by the organizations that offer training? We provide unique evidence on these questions by combining a household census (N=42,895) in urban Bangladesh, a randomized controlled trial of an on-the-job training intervention, unique data on the targeting of the implementing partner, and a structural model. We find that take-up of the training at full price is minimal, and there are no effects on labor market outcomes. The poor, however, increase employment and consumption under a 70% discount. We also document that, for the same program, the implementing NGO would have selected richer beneficiaries who are more likely to enroll in training. Our structural model rationalizes these findings by estimating that the poor have higher-than-average returns to training, but they also experience higher participation costs, and are hence targeted less frequently by an implementer that wants to maximise take-up. We confirm the central finding that the poor have higher-than-average returns to training in a re-analysis of previous training experiments, using machine learning predictions of outcomes and treatment effects.

Development Innovation Ventures

JPAL-KCAI, Gates Foundation

The bulk of the world's extreme poor work in subsistence agriculture. Diversification out of this activity is often seen as the sine qua non of economic development. We evaluate whether the roll-out of a mainstay development intervention—microfinance—into poor, agricultural and largely unbanked populations in rural Uganda helps borrowers to diversify into non-agricultural labour activities. The new microfinance product is targeted to women, and differs from existing sources of formal and informal credit in that it allows them to borrow larger amounts but has inflexible repayment dates and the use of funds is monitored. We find that the arrival of microfinance enables women to diversify out of agriculture and into service-based activities such as small-scale trading. This low-level structural change, however, is not transformative in that it does not lead—at least after two years—to significant uplifts in earnings, consumption, savings, investment and overall wealth.