Research

Information asymmetries make contracting difficult within firms. Standard principal-agent theory predicts that monitoring should unambiguously raise workers’ effort. I test this prediction using a field experiment with Liberian trucking companies. Results show that drivers who are experimentally monitored provide more effort on average. However, a key finding is that this effect is heterogeneous across drivers: some drivers provide less effort as a result of monitoring and managers are therefore reluctant to monitor them. I discuss why this might be and show that decreasing effort may be a way for drivers to retaliate against the manager’s decision to monitor them.

Link to Draft

Coverage: VoxDev Podcast

Small business owners in low-income settings often face information frictions and limited access to professional networks. While formal business training programs have been widely studied, less is known about whether low-cost virtual peer interactions can facilitate the diffusion of business knowledge and practices. We study a randomized experiment in Liberia involving 1,131 micro-entrepreneurs, where the treatment group joined small, moderated discussion groups that met weekly by phone to discuss business challenges. We find that participation in virtual discussion groups leads to significant changes in business behavior. Treated entrepreneurs adopt more innovative practices, including new marketing strategies and sales locations; exhibit higher levels of financial planning; increase their use of digital technologies such as mobile phones, mobile money, and social media for business purposes; and reorient their advice-seeking toward other business owners. A distinctive feature of the study is the availability of detailed records of discussion content. We link treatment effects to the topics most frequently discussed within groups. Outcomes that are more commonly discussed—such as marketing strategies, digital tools, and saving practices—exhibit larger treatment effects, providing descriptive evidence on peer learning mechanisms. In contrast, we find no significant average effects on short-term profits or revenues, consistent with learning-driven adjustments that may precede performance gains. We document heterogeneity, with revenue gains concentrated among entrepreneurs who plausibly faced weaker access to high-quality business networks prior to the intervention. Together, the results show that virtual peer networks can shape business practices and information flows in settings where formal institutions and networks are limited.

Link to Paper

Access to capital is widely believed to enable small businesses to grow and create jobs, but the general equilibrium effects of such interventions remain poorly understood. We evaluate the market-level impacts of a large cash grant program offered to microenterprise owners in Liberia. Using a two-stage randomized design—first randomly assigning half of markets to treatment and then allocating grants by public lottery within treated markets—we identify both direct effects on recipients and spillover effects on untreated competitors operating in the same market. We find that grant recipients experience substantial increases in revenues, savings, investment in productive assets, and employment. Strikingly, untreated firms in treated markets also record higher sales and profits relative to firms in untreated markets, with smaller employment gains and little change in assets. This pattern is most consistent with spillovers driven by local demand expansion rather than business stealing or input cost reductions. Our findings suggest that the social returns to such interventions may exceed the private returns, with important implications for policies aimed at firm growth and local economic development.

This paper experimentally tests two distinct strategies to alter how drivers apprehend risk on the road. One first strategy mimics the typical information-provision intervention and involves providing drivers with information on road safety risks and the role of driver behavior in preventing such risks. Surprisingly, this approach shows no significant effect on driver beliefs and a moderate impact on behavior. In contrast, the second strategy draws from the motivated reasoning literature, aiming to alter drivers’ beliefs and behaviors by addressing their underlying preferences. The results provide evidence that drivers’ beliefs about risk are shaped by their preferences over their driving activity. This study shows that in order to change beliefs it can be more effective to target preferences than beliefs directly, and opens up innovative avenues for developing road safety interventions.

Link to Paper

Emerging market economies are claiming a rising share of trade in products associated with the new technologies that help reduce pollution, limit greenhouse gas emissions, and help mitigate the impact of extreme weather. Indeed, technological advances and falling costs in areas such as renewable energy, are helping developing economies close in on high-income countries’ dominance of the sector, new research from IFC shows. To continue to capitalize on opportunities in this area, low- and middle-income countries will need the help of sound policy making and increased investment in research, presenting a key opportunity to international investors. 

Link to report

Follow-up analysis: “Green innovation: opportunities for developing countries”, with Ralf Martin and Trang Thu Tran, Chapter for Pathways to a productive and inclusive net zero, February 2026