Current Projects


Lead is a very toxic pollutant; even trace amounts can affect the cardiovascular, reproductive, and neurological systems. Much of what we know about the costs of lead pollution come from rich countries, but today the vast majority of lead pollution exposure occurs in middle- and low-income countries. In our study we look at the impact of industrial lead pollution on test scores in Mexico. To overcome the lack of lead pollution monitoring data, we take advantage of a 2009 policy change in the United States that lead to increased used lead acid battery (ULAB) recycling activity in Mexico. We use a difference-in-difference approach to estimate the impact of this ramp up of lead intensive industrial activity, comparing Spanish and mathematics standardized test scores before and after this 2009 increase in industrial activity between students who attended school closer to a ULAB-recycling facility and those a bit farther away. We find that the increase caused a 0.1 standard deviation decrease in test scores, with effects stronger for mathematics. These are large effect sizes, equivalent, but opposite sign, to the effects of many education policies, like teacher training, which are aimed at improving learning outcomes.


In this project, we undertake an impact evaluation of Senegal's  Programme National de Biogaz Domestique (PNB-SN), or the national domestic biogas program. PNB-SN aims to support Senegal's energy transition by promoting household use of biogas to meet energy needs. Specifically, the program seeks to reduce the use of solid fuels, like firewood, for cooking and improve agriculture production through the use of bioslurry fertilizer, a bi-product of anaerobic digestion. Using data from household surveys conducted in 2019, we evaluate the impact of PNB-SN on a wide array of household and individual-level outcomes. 


This paper reports on the universe of garment-making firm owners in a Ghanaian district capital during the COVID-19 crisis. By July 2020, 80% of both male- and female-owned firms were operational. Using pre-pandemic data, we document that selection into persistent closure differs by gender. Consistent with a "cleansing effect" of recessions and highlighting the presence of marginal female entrepreneurs, female-owned firms that remain closed past the spring lockdown are negatively selected on pre-pandemic sales. The pre-pandemic sales distributions of female survivors and non-survivors are significantly different from each other. Female owners of non-operational firms exit to non-employment and experience large decreases in overall earnings. Persistently-closed male-owned firms are not selected on pre-pandemic firm characteristics. Instead, their owners are 36 percentage points more likely to have another income generating activity prior to the crisis and fully compensate for revenue losses in their core businesses with these alternative income generating activities.

Abstract | Policy Brief

Energy has been called the “golden thread” that connects economic growth, social equity and environmental sustainability, but important knowledge gaps exist on the impacts of low- and middle-income country energy interventions and transitions. This study offers perhaps the broadest characterization to date of the patterns and consistency in quantitative and peer-reviewed social science literature considering such impacts. Starting from approximately 80,000 papers identified using a search procedure organized along energy services, technology, and impact dimensions, and structured to achieve breadth and replicability, articles were first screened to yield a relevant subset of 3,000 quantitative papers. Relevance is defined as providing one or more types of impacts on intra-household, household, firm, public service, national economy, or environmental outcomes. A set of heat maps highlights areas of concentration in the literature, namely work that emphasizes the negative health and pollution effects of traditional cooking and fossil fuel use. The extent and consistency of evidence for different types of impacts (in terms of direction and statistical significance) is also discussed, which reveals considerable heterogeneity and highlights important knowledge gaps that remain despite rapidly expanding energy scholarship. The patterns of evidence are also surprisingly consistent across methods. The article concludes by articulating several research challenges that should motivate current and future generations of energy and development scholars.


Rural electrification (RE) is a core component of the Sustainable Development Goals and a major focal point of the global development community. Despite this focus, more than one billion people worldwide lack access to electricity, and electrification rates need to more than quadruple to meet international goals. We believe that lack of progress is partly driven by a know-do gap, a misalignment between academic research and the information needs of policy makers. Most studies measuring the impacts of electrification focus on precise estimation of a few outcomes, specifically health, education and productivity impacts. Other important impacts, e.g. environmental, have remained largely unstudied. As a consequence, quantifying the full set of costs and benefits of expanding electricity access is difficult and rarely done. When cost benefit analyses are done, they are often incomplete, and conclusions are highly susceptible to unavailable or uncertain parameters. We illustrate these arguments in the case of Bhutan, where RE rates have expanded rapidly in the past few decades. We show that RE via grid extension had positive impacts related to fuelwood consumption, education, and employment, but we do not find an effect on health. We then use these impact estimates to conduct cost-benefit analyses. For the cost-benefit parameters not available from our impact evaluation, we transfer reasonable estimates from related contexts. To acknowledge the uncertainty induced by this process, we conduct Monte Carlo analyses and confirm that, while the private NPV calculations are robust to alternative parameter values, the social returns are sensitive to estimates of the social cost of carbon and costs of grid operation and maintenance. Based on this exercise, we highlight research gaps that persist and that preclude 1) careful cost-benefit analysis of RE more generally and 2) financial investment in the sector.


In this chapter in the Handbook of Environmental Economics (Volume 4) we examine human–environmental interactions and their implications for health, highlighting the ways in which economic tools provide analytical insight into this relationship as well as potential solutions to preventing (and treating) diseases. Specifically, we contend that the household production framework – which focuses on the beneficiary and households – provides a helpful conceptual tool to understand when and how households will expend resources to avert environmental risks. We review the empirical literature in environmental health economics, focusing particularly on studies that value environmental risk reduction, that examine household adoption of environmental health technologies, and that evaluate health impacts of these technologies. While economists have been mostly on the sidelines of the research on environmental health in LMICs, there is a growing empirical literature on values, drivers, and impacts. This literature reveals considerable heterogeneity in estimates and some methodological blind spots, with sobering overall findings. While there are exceptions to the following observations, we find relatively low values for environmental risk reductions, which is mirrored by limited adoption of environmental health technologies and, accordingly, disappointing results regarding health impacts. Because much remains to be learned about valuation, evaluation, and adoption, we conclude by outlining a research agenda for LMICs that must focus on multiple risks, on the political economy of how interventions are supplied, and on environmental disasters, especially in the light of mounting climate change risks.