This dissertation analyzes how poor households in Africa and South Asia respond to large-scale policy experiments involving conditional and unconditional cash transfers as well as interventions that have training and behavioral components.

The first chapter, titled “Do Cash Transfers Improve Women’s Agency? Evidence from Lab and Field Experiments", details results from lab-in-the-field experiments that I have designed and implemented for my PhD job market paper. This chapter investigates whether receiving a cash transfer - in a very low-empowerment context - improves women’s agency in household decision-making, particularly, in the longer term. I try to answer this question by combining data from a female-targeted unconditional cash transfer (UCT) program with lab- in-the-field experiments. The UCT was designed as a Randomized Controlled Trial and was disbursed to adult women from ultra-poor households in rural Nigeria over a period of 15 months. The lab-in-the-field experiments were carried out on 503 married couples one year after the UCT program ended and can measure women’s (as well as men’s) agency in household decision-making. The main experimental measure of agency is the propensity to not defer decision-making to one’s spouse. By randomly varying whether subjects’ decisions can be observed by their spouses, a latent desire for agency (in the absence of spousal observation) versus an actual manifestation of one’s agency (when decisions are observed by one’s spouse) can be separately measured. I find that UCT-receiving women, one year after the completion of the program, defer 13 percentage point fewer decisions to their spouses; however, they do so only when their decisions cannot be observed by their spouses.

In the second chapter, titled “Social and Financial Incentives for Overcoming a Collective Action Problem: Sanitation Underinvestment in Rural Bangladesh", we study the effects of social and financial incentives on communities’ ability to overcome collective action problems. Our specific context is a sample of 107 villages (approximately 19,000 households) in rural Bangladesh, and the collective action problem we study is investment in hygienic latrines and their subsequent maintenance and use. We randomized (1) whether and what type of incentive was provided - a financial reward (which was, essentially, a conditional cash transfer) or a non-financial "social recognition" reward, and (2) whether and what type of verbal commitment the households were encouraged to make - a private pledge vs. a public pledge. We measure short-term (3 months) and medium-term (15 months) effects and investigate the mechanisms behind the effects. Preliminary reduced form estimates suggest that a small financial reward has the strongest impact in the short term, inducing an approximately 12 percentage point increase in hygienic latrine ownership relative to ‘control’ villages, which received basic health messages. The public commitment treatment, in which members would publicly commit to have a hygienic latrine at the end of a group meeting, induced an approximately 5 percentage point increase in hygienic latrine ownership in the short term. In the medium term the effect of the financial reward dissipates; however, the smaller effect of the public commitment treatment in the short term continues to persist after 15 months. The results suggest that low-cost interventions which take advantage of social network dynamics have the potential to improve the effectiveness of group-meeting based rural health interventions common in developing countries.

The third chapter, titled “Training Mentors? Experimental Evidence from Female-Owned Microenterprises in Ethiopia" reports findings from a randomized evaluation of a business training and mentoring intervention targeted at female- owned microenterprises in Ethiopia shows that formal business training produces an immediate impact on the adoption of beneficial business practices (that were highlighted in the training); however, no impact on business profit is observed in the short term. Three and four years after the training, we observe delayed impacts on business profits. Shortly after the training, the trained cohort is randomly assigned to provide mentoring to less-experienced women who own smaller businesses. These mentees were nominated by mentors at baseline from their own social networks. The overall impact of mentoring on mentees is more muted. There are early impacts on the adoption of beneficial business practices, and some measures of profits. However, the impacts on profits do not persist in the longer term.