Chapter 1: Western stakeholders are increasingly demanding that multinationals sourcing from developing countries be accountable for labor rights and working conditions upstream in their supply chains. In response, many multinationals privately enforce labor standards in these countries, but the effects of their interventions on local firms and workers are unknown. I partnered with a set of multinational retail and apparel firms to enforce local labor laws on their suppliers in Bangladesh. I implemented a randomized controlled trial with 84 Bangladeshi garment factories, randomly enforcing a mandate for worker-manager safety committees in 41 supplier establishments. The intervention significantly improves compliance with the labor law. It also has a small, positive effect on indicators of safety committees' effectiveness, including measures of physical safety and awareness. Factories with better managerial practices drive these improvements. In contrast, factories with poor managerial practices do not improve compliance or safety, and in these factories, workers' job satisfaction declines.

Chapter 2 (joint with Rachel Heath and Tyler McCormick): Many workers in large factories in developing countries are internal migrants from rural areas. In collaboration with Rachel Heath and Tyler McCormick, I examine the relationship between workers' migration status and the working conditions they face in a household survey of garment workers in Bangladesh. We document that migrants are in firms with higher wages but worse working conditions, but as their careers progress, they have higher mobility than locals as they move towards firms with better conditions. These facts are consistent with a model in which migrants are poorly informed about working conditions upon beginning work but learn more as they gain experience in the industry.

Chapter 3: I test for the presence of compensating wage differentials for factory building safety in Bangladesh's apparel sector. I find no evidence in support of a compensating wage differential for building safety. This descriptive fact is not explained by heterogeneity in workers' or factories' observable characteristics. Instead, I show that workers have incomplete information about factories' compliance with building safety standards, which I argue is difficult for workers to observe. Workers at high compliance factories underestimate their factories' performance on building safety audits, while workers at low compliance factories dramatically overestimate their factories' performance. I implement a pilot field experiment with 308 garments workers in which I randomly intervene to provide workers with information about their factories performance on the audits relative to other factories nearby. The information causes workers to correctly update their beliefs about their factories' safety. It reduces turnover among workers at high compliance factories. Among workers at low compliance factories, turnover is unaffected, but workers are less likely to make referrals to their factory.