Governments and NGOs worldwide subsidize the adoption of beneficial household technologies. However, these subsidies may be mistargeted when the benefits depend on continued use, which is not guaranteed by initial take-up. I study this misallocation issue in Mexico City’s Rainwater Harvesting (RWH) Program, which distributes 10,000 free RWH systems annually on a first-come, first-served basis. Using program evaluation methods and instrumenting program participation with contingent valuation responses, I estimate treatment effects across the distribution of unobserved willingness to pay (WTP). I find that households’ ex-ante WTP for the technology predicts its ex-post usage and benefits, controlling for sociodemographics. High-WTP households save 5 hours per week on water procurement and reduce the postponement of daily activities due to water scarcity by 25 pp. Conversely, low-WTP households are less likely to use the technology, resulting in negligible benefits. Counterfactual simulations indicate that households with a usage rate 20 pp lower than the average would be discouraged from taking up the technology if the subsidy rate were reduced from 100% to around 97%. This straightforward approach can inform subsidy calibration when incentivized experiments and randomized interventions are not feasible.   

This paper describes how urban households secure water when piped networks are unreliable, using a representative sample of 2,197 households in Hargeisa, Somaliland —a rapidly growing, semi-arid city on the frontier of climate stress. Rather than relying on a single source, households assemble “patchwork” portfolios that combine piped connections, tanker trucks, rainwater harvesting, and other sources, adjusting their composition across the dry and rainy seasons. Four findings emerge: (1) source portfolios vary widely across households and seasons; (2) unconnected households pay more and face more shortages; (3) costs and access are largely regressive; and (4) storage is a critical coping asset that poorer households cannot afford. Planned network expansion will narrow, but not eliminate, these inequalities.

Increasing block pricing schemes represent difficulties for applied researchers who try to recover demand parameters, in particular, price and income elasticities. The Mexican residential electricity tariff structure is amongst the most intricate around the globe. In this paper, we estimate the residential electricity demand and use the corresponding structural parameter estimates to simulate an energy efficiency improvement scenario, as suggested by the Energy Transition Law of December 2015. The simulated program consists of a massive replacement of electric appliances (air conditioners, fans, refrigerators, washing machines, and lights) for more energy-efficient units. The main empirical findings are the following: in the main counterfactual scenario, the overall residential electricity consumption decreases 9.9% and the associated expenditure falls 11.3%. Additionally, the electricity subsidy decreases 7.5 billion of Mexican Pesos per year (i.e., 403 million of USD at the average exchange rate registered in 2017) and there is an annual cut in CO2 emissions of 3.9 million of tons. 

Water scarcity is an increasingly pressing issue, prompting utilities and regulators to explore cost-effective strategies to reduce residential water consumption. While price-based policies are commonly used, empirical evidence suggests that they often have limited impact. In collaboration with a public water utility in Colorado, we designed and implemented a randomized controlled trial to assess the effectiveness of informational interventions in reducing inefficient lawn irrigation practices. Using real-time data from advanced metering infrastructure (AMI), we identified irrigation episodes among 7,000 households during the summer of 2024. Our intervention delivered targeted alerts to customers who watered their lawns during inefficient daytime hours or exceeded the recommended irrigation frequency. We find that the intervention resulted in a 5% to10% reduction in daytime irrigation events and a decrease in weekly irrigation frequency, depending on the treatment. Our preliminary evidence also indicates that total water use decreased by 4% among households that received alerts about irrigating too frequently, while remaining largely unchanged among those that received alerts about irrigation timing.

We examine whether improving maintenance practices for existing domestic Rainwater Harvesting Systems can reduce household water insecurity among owners and whether subsidies for rainwater collection can decrease water use from overburdened sources. This full-scale RCT builds on two pilots conducted in 2023–2024, which indicate that poor maintenance may be causing households to collect significantly less water than these systems can deliver under ideal conditions. Our experiment, scheduled for 2027 in Mexico City, will randomly assign 650 households to one of three groups: hands-on training, professional maintenance, or a control group. Mid-season, half of the households will be offered a per-liter reward for the rainwater they collect. Our main outcome is the volume of rainwater collected, supplemented by water quality measures to explore behavioral drivers and perceived value. We will also assess impacts on time use and expenses. Outcomes will be measured using water meters installed by our team, at-home quality tests, and surveys. Our findings will evaluate the cost-effectiveness of strategies that lower barriers to effective system use, increasing benefits for households and returns for the city. Additionally, we will determine whether subsidizing sustainable sources can promote substitution away from overused sources as an alternative to politically difficult and often ineffective price changes.

Boreal forest fires are a large and underpriced source of carbon emissions. Yet the vast, remote zones where these fires burn are largely left unmanaged because timely detection is infeasible. We quantify the climate value of closing this detection gap, using the FireSat satellite constellation as a concrete case. Combining nineteen years of spatially explicit fire and carbon data from the NASA ABoVE Fire Emission Database with the value-of-information and cost-plus-net-value-change economic frameworks, we estimate the carbon emissions avoided when early detection allows remote fires to be attacked before they escape, and value these avoided emissions at the social cost of carbon. Focusing on the permafrost-rich ecoregions of Alaska and the Yukon, where suppression yields an unambiguous net climate benefit, we find that early attack avoids emissions at a cost of approximately $2.61–8.30 per tonne of CO2, implying a benefit-to-cost ratio of 22–71 to one even after accounting for additional suppression costs. Our estimates are conservative: they exclude fire-induced permafrost-thaw emissions and value only avoided carbon, omitting air-quality, health, and property co-benefits. Our results suggest that early wildfire detection in carbon-rich boreal landscapes is a highly cost-effective lever for climate mitigation.

Although forced displacement disproportionately affects low- and middle-income countries, where nearly 75 percent of the world’s refugees live, the bulk of relevant data currently comes from high-income countries. This project built on recent World Bank efforts to collect representative data on forcibly displaced peoples and their hosts in several countries to harmonize representative surveys covering ten countries that hosted displaced people in the 2015-2020 period. 

The resulting harmonized database can be accessed here. 

The findings from the harmonized surveys are summarized in three Policy Briefs.