Y-RISE — RCT of Water Entrepreneurship Program Grant (July 2024)

Note: This page summarizes the rationale behind a GiveWell grant to the Yale Research Initiative on Innovation and Scale (Y-RISE). Y-RISE staff reviewed this page prior to publication.

Published: October 2024

The organizations

The Yale Research Initiative on Innovation and Scale (Y-RISE) identifies and tests promising health and livelihood interventions, conducts research on the challenges and implications of scale as interventions grow, and works to build the field of researchers and implementers interested in the science of scaling.1

Y-RISE began in 2017, and its main model focuses on working with affiliated researchers and institutions to examine the complexities of taking a development project to scale.2 It has worked on projects including incentives for migration, COVID-19 vaccine uptake, vaccination programs, workforce development, electrification efforts, mental health initiatives, climate resilience interventions, and more.3

In December 2022, we recommended a $1.35m grant to Y-RISE to support its core operations.

BRAC is an international development organization, founded in Bangladesh. It has scaled to reach more than 100 million people worldwide.4 BRAC is well-known for its microfinance, social enterprises, and ultra-poor graduation programs, but operates across a number of other areas, including water, sanitation, and hygiene, education, and climate change.

The intervention

Coastal regions of south and south-east Asia have seen rising sea levels, tidal flooding, and storm surges. This has led to the contamination of freshwater sources with saltwater, creating a shortage of water that is safe to drink. In coastal Bangladesh, Y-RISE estimates that water salinity levels are three-to-five times higher on average than is considered safe for human consumption.5 High salinity is associated with increased hypertension, a major risk factor for cardiovascular disease, and, for pregnant women, with increased preeclampsia, a major risk factor for maternal mortality.6

In addition to salinity, arsenic contamination is a major problem with Bangladesh’s water supply. One estimate suggests that 57 million people in Bangladesh have been chronically exposed to drinking water with arsenic levels exceeding WHO standards.7 According to the WHO, long-term effects of arsenic exposure include skin lesions; cancers of the skin, bladder, and lungs; pulmonary and cardiovascular disease; adverse pregnancy outcomes; and negative effects on cognitive development.8

BRAC’s program is designed to increase the adoption of reverse osmosis (RO) water treatment technology.9 Much like chlorine, RO equipment addresses pathogen contamination. Unlike chlorine, it also removes sodium, magnesium, lead, and arsenic, which means, if properly deployed, it is able to address the issues with Bangladesh’s water supply outlined above.10

RO technology is relatively expensive on a per-unit basis, which makes it impractical for household-level implementation.11 In light of this, BRAC’s program identifies and trains local, profit-motivated entrepreneurs who are willing to invest in RO to treat and then sell the desalinated water to their local communities. BRAC Microfinance provides loans of up to $10,000 to the entrepreneurs, covering the purchase of the RO equipment, land, and installation. Through this scheme, the entrepreneurs are able to provide between 500 liters and 1500 liters of treated water per hour, equivalent to the drinking and cooking needs of between 200 and 600 households.12

So far, BRAC has implemented a small-scale pilot of this program with 360 water entrepreneurs, which has shown promising loan repayment (>90%) and business success (96%) rates.13

The grant

This grant will fund Y-RISE to conduct a cluster randomized controlled trial (RCT) of BRAC’s desalination entrepreneurship program in coastal Bangladesh.14 The grant covers the research costs for the trial. BRAC has committed to covering the cost of program implementation.15

The trial will randomly assign 365 communities (totalling 10,950 households) to either a treatment group (240 communities) or control group (125 communities).16 Within the treatment group, half of the communities will be served by two water entrepreneurs, while the other half will be served by one. Within both treatment groups, Y-RISE will randomly assign two further interventions: (a) the facilitation of a water delivery business (delivering cans of desalinated water door-to-door to paying customers), and (b) the provision of discount coupons, for redemption at entrepreneurs’ kiosks, allowing the researchers to estimate how much water consumption changes when prices change. BRAC will compensate the entrepreneurs for the coupons’ value.17

• The number of households who purchase or receive clean water from the kiosks, and how this varies by subsidy level;
• The quality of water consumed by the households, as measured by (a) microbiological contamination levels, (b) salinity levels, and (c) arsenic levels;
• Health outcomes, including (a) the prevalence of diarrheal disease, (b) the prevalence of preeclampsia in pregnant women, (c) anthropometric outcomes in children, and (d) mortality, though the study will not be powered to detect a statistically significant mortality effect;
• The time spent collecting drinking water;
• Medical costs averted;
• The income benefits of the program for both water entrepreneurs and for households;
• The price of water;
• Program costs, including (a) the cost per household reached, (b) the number of households per desalination kiosk catchment area, and (c) the cost of vouchers for ultra-poor households.

Budget and timeline

The budget for the grant is $1,299,146.21 This is broken down as follows:

• $603,889 to a survey firm22
• $409,035 to Yale for personnel
• $22,000 to Yale for travel expenses
• $146,119 to Yale for other direct costs, including household and kiosk microbiological water testing
• $62,715 to Yale for indirect costs

This budget covers two years and nine months of fieldwork time. This includes six months for preparation work and piloting, one year for baseline data collection, and one year for 12-month follow-up.23 A three month buffer was added in case of data collection overruns.

We expect the preparation work to begin immediately following the award of this grant and conclude by December 2024. This period will involve piloting the intervention and survey instruments, and preparing the sample frame based on mapping of water salinity and scarcity levels in the region.24

We expect baseline data collection to occur in February and March 2025, with endline data collection in February and March 2026. Baseline and endline data collection will include household water quality tests. In addition, between December 2025 and May 2026, the study will include monthly follow-up surveys with the entrepreneurs to test the quality of their water. There will be a 12-month follow-up survey in February and March 2027.

How will we know if the grant was successful?

We view this grant as an opportunity to learn about a water quality intervention that is new to GiveWell, and to experiment with supporting a study that has the explicit goal of influencing the funding decisions of another organization. Since learning value is an explicit focus of this grant, we have defined in advance a range of short-term and long-term indicators of grant success, which we plan to use to assess whether to consider similar grants in future.

We will judge the success of this grant in the short term (i.e., within the grant period) by assessing whether:

• The evaluation is being implemented in a manner consistent with our expectations (e.g. if random assignment is successful, the research team pre-registers a pre-analysis plan, and there is general adherence to the evaluation timeline);
• Take-up of the technology is high enough for a successful trial;
• BRAC seems to have bought into the program (according to our subjective assessment and BRAC’s continued funding of the program), in both its Climate Change and Microfinance teams.

We’ll judge the success of this grant in the long term (i.e., after the grant period) by assessing whether:

• The trial was well-conducted (e.g. the study arms were well-balanced at baseline, there was a low attrition rate across the study period, outcomes were measured successfully, and the research team adhered to the pre-analysis plan);
• We have a credible signal that the results of the study informed BRAC’s programming. This could involve BRAC scaling the program up in the event of positive results, or declining to do so if the study has a null result;
• The grant leads to ongoing collaboration between GiveWell and BRAC, and results in our scoping further grant opportunities with BRAC in the future.

The case for the grant

We are recommending this grant because:

• This grant appears highly cost-effective, due to the possibility it could increase the cost-effectiveness of a large amount of BRAC’s programming (most important factor). Our best guess is that this benefit stream is equivalent to 18 times as cost-effective as cash transfers.25
  • The intuition here is that the study could cause BRAC to direct money toward more cost-effective programs than it would otherwise fund, if this evaluation finds that the water entrepreneur model yields health or economic benefits. Specifically, we believe that BRAC has the willingness and capacity to scale up the program to all 20 million people currently affected by saltwater intrusion in coastal Bangladesh.26
  • BRAC’s Microfinance arm serves 11 million clients, and disbursed $6 billion in microloans in Bangladesh in 2023.27 We assume that these programs are about as cost-effective as cash transfers. If our current estimate of the cost-effectiveness of this desalination program is approximately correct, then this grant would clear our 10x cost-effectiveness bar if it has just an 18% chance of affecting approximately $60 million of BRAC Microloan spending.
  • This path to impact seems plausible because:
    • BRAC has already committed up to $2 million in loans to water entrepreneurs as part of the trial, so is invested in its results;28
    • We spoke to BRAC’s Associate Director of Microfinance, Amit Kanti Sarker, and believe that BRAC is excited about the potential for the study to inform a potential scale-up of the program;29
    • Y-RISE has a partnership with BRAC that is explicitly focused on research to inform BRAC’s programming;30
    • BRAC’s Climate Change team is also working closely with Y-RISE on program design, targeting, training, and overseeing implementation.31 The grant may therefore also influence BRAC’s Climate Change team (as well as its Microfinance team), though our cost-effectiveness analysis has focused on the latter.
• This grant will allow us to learn more about the cost-effectiveness of desalination in Bangladesh, and it’s possible that we learn the program is above our cost-effectiveness bar. We model this ‘value of information’ from this grant as 8 times as cost-effective cash transfers, separate from the leverage benefit stream above.
  • Although we currently think it is likely that BRAC’s program is less than ten times as cost-effective as cash transfers, our estimate is highly speculative. A number of factors could lead us to believe the program is above our funding bar. For example:
    • We could be underestimating the reduction in hypertension resulting from desalinated drinking water;
    • Our current model could be missing a health benefit stream, e.g. by reducing arsenic contamination;
    • Our current model could be missing an economic benefit stream, e.g. in reducing the time spent collecting water, or in preventing productivity losses from illness caused by contaminated water;
    • We could be overestimating the program’s costs, or;
    • There could be an opportunity to fund a specific, cost-effective portion of the program, such as vouchers to cover the cost of desalinated water for the ultra-poor.32
  • This RCT will measure a number of exploratory outcomes that will help us get clearer on the most important benefits of the program, and address some of the factors above.33 If we update positively on these factors, and the program looks cost-effective by GiveWell’s criteria, then this would present an opportunity for GiveWell to recommend additional, cost-effective grants towards desalination programs in the future.
  • This grant will also help us learn about parameters in our cost-effectiveness models that are useful for our grantmaking in water quality beyond desalination, such as by providing a direct estimate of medical costs averted.
• We believe that Mushfiq Mobarak, the lead researcher of this RCT and the founder and faculty director of Y-RISE, is highly aligned with GiveWell’s values, and conducts high-quality RCTs. Professor Mobarak is on GiveWell’s Research Council, and has a longstanding relationship with BRAC. Examples of high-quality RCTs with Professor Mobarak’s involvement include a 2021 study on mask use, a 2024 study on vaccine uptake in Sierra Leone, and a 2017 study on the effects of emigration on rural labor markets.
• This grant presents an opportunity to develop a closer relationship with BRAC, which could allow us to identify other cost-effective grantmaking opportunities in the future. We think BRAC is a promising candidate for future GiveWell funding, due to its large geographic footprint and extensive community health worker network. We hope to get a better sense for where cost-effective grant opportunities with BRAC might lie as a result of our collaboration on this grant.
• This grant fits within GiveWell’s Water grantmaking team’s research strategy. One of the team’s goals this year is to identify opportunistic, light-touch ways to explore new areas and interventions within the sector, and to broaden our grantmaking focus beyond chlorination.

Risks and reservations

Our main reservations about this grant are:

• The evaluation might not change BRAC’s programming. BRAC has seen high repayment rates (>90%) in its pilot of the program. This might incentivize BRAC Microfinance to provide these loans regardless of whether the trial finds the program to have health or economic benefits. However, when we spoke to BRAC about this, it said it saw its purpose as supporting programs that improve welfare rather than solely focusing on loan repayment. Our impression is that BRAC is only committed to scaling the program if the study finds positive effects on health and economic or financial outcomes.34
• The program might not be cost-effective. At present, we speculatively model the program at between 2 and 9 times cash transfers. This estimate should be treated with caution, as it is not informed by reliable estimates of costs, or the effect of desalination on cardiovascular or under-five mortality. However, we think it conveys two reasons to be skeptical about the relative cost-effectiveness of desalination programs: (a) compared to water chlorination, where we have focussed our grantmaking in water in the past, RO is more expensive to deliver, due to high capital costs35 ; and (b) relative to sub-Saharan Africa, our general understanding is that the all-cause mortality burden tends to be lower in the areas where desalination may be most important, like Vietnam, Myanmar, or India. Notwithstanding these concerns, we think it is plausible that if, for example, GiveWell covered a smaller share of program costs (such as water subsidies specifically for poorer households, pregnant women, or households with children under five), or if we were to receive a positive update on this program through the RCT, then desalination could still meet our funding bar.
• Desalination may have limited room for further funding above GiveWell’s cost-effectiveness bar. If the regions most affected by rising salinity levels have relatively low baseline mortality, then even if the study finds that the program leads to meaningful improvements in health outcomes, it may still not look as cost-effective in GiveWell’s terms as programs targeting areas with higher mortality burdens. However, we think desalination programs could potentially be targeted to areas with relatively higher baseline mortality, particularly over time, as climate change and saltwater intrusion become more acute problems in coastal regions.
• Desalination may have adverse effects. First, if discharged improperly, brine, a byproduct of desalination, could increase the salinity levels of aquifers or soil areas. To mitigate this risk, Y-RISE plans to include a protocol for the safe discharge of brine, and to investigate alternative uses of brine that do not require discharge into the local environment.36 Second, our understanding is that reverse osmosis is energy-intensive, which could drive up the cost of the program, or serve as a drain on the country’s energy infrastructure.37 We expect to learn more about this risk through the study’s cost analysis.

Plans for follow up

Following the award of this grant, we plan to request and review a draft pre-analysis plan from the researchers, to ensure we’re on the same page on the study design. Following this, we’ll check in with the research team approximately every six months, aligning our check-ins around key study milestones (like registering the pre-analysis plan, analyzing the baseline survey of households and water entrepreneurs, launching random assignment, starting the intervention, wrapping up endline surveys, and conducting endline analysis).

We also hope to establish a closer relationship with the relevant program teams at BRAC across the grant period.

Internal forecasts

For this grant, we are recording the following forecasts:

Our process

Given the small size of this grant, our investigation was relatively light touch:

• Y-RISE brought this study to our attention during a regular check-in conversation in late 2023;
• We spoke to Y-RISE about this grant in detail in February 2024, and received a full proposal in April 2024;
• Following receipt of the proposal, we spoke to BRAC, and had extensive written correspondence with Y-RISE on the details of the planned evaluation;
• We put together a back-of-the-envelope calculation in May 2024 for (a) direct delivery of the water entrepreneur model, (b) the value of information of the RCT, and (c) the cost-effectiveness of leveraging BRAC’s funding towards this program in the event the study shows it to be successful.

Sources

• 1
  • “Y-RISE works across the various stages in scaling a program, from ideation and design to implementation, providing research, expertise, data management, and coordination. It also holds an annual conference for its network affiliates and partners to explore the complexities of scaling.” GiveWell's non-verbatim summary of a conversation with Y-RISE, November 14, 2022, p. 1.
  • "Y-RISE teams work through research projects in collaboration with implementing organizations to outline scaling complexities… At Y-RISE however, we believe that scaling research should not involve a trade-off between rigorous research and a delay in implementation: accordingly, we are focused on forming long-term partnerships with institutions and implementing organizations to iterate on and continuously improve programs as they scale. Finally, we engage in network building as we believe that there is a need to build the field of scaling research as an area of scientific inquiry, both for researchers and for practitioners, to ensure the right collaborations and skill sets." Y-RISE, Budget narrative, 2022, p. 1-2.

• 2

  "The Yale Research Initiative on Innovation and Scale (Y-RISE) is likely the only center dedicated entirely to studying the complexities of scaling interventions. One of its goals is to generate new research on scaling and make it useful for policymakers. It has created a network of academic researchers to engage in this research. … Systemic change requires long-term partnerships with sophisticated institutions that work in multiple countries. Y-RISE has two major partnerships, one with BRAC and the other with the World Health Organization (WHO)." GiveWell's non-verbatim summary of a conversation with Y-RISE, November 14, 2022, pp. 1, 5.

• 3

  "Y-RISE and affiliated researchers have engaged in research on a number of programs, including:

  • Migration incentives for seasonal poverty
  • COVID-19 vaccine uptake in Sierra Leone
  • Replicating and scaling "access first" vaccination programs
  • Workforce development
  • Electrification efforts

  These projects are in various stages—some are funded and moving forward, others are in initial implementation stages, and in other cases Y-RISE is talking with potential implementation partners. Other research projects are in initial scoping and proposal phases. These include:

  • Mental health initiatives
  • Climate resilience interventions
  • Healthy nutrition in schools in Jamaica
  • Bunded vaccination in Sierra Leone
  • Electrification outages advocacy

  Y-RISE is supporting research in many areas, understanding that some of those efforts are likely to result in opportunities for scaling, while others are likely to show that the programs are not likely to be effective at scale." GiveWell's non-verbatim summary of a conversation with Y-RISE, November 14, 2022, pp. 2-3.

• 4

  “BRAC is an international development organisation founded in Bangladesh in 1972 that partners with over 100 million people living with inequality and poverty to create opportunities to realise human potential.” BRAC, Who we are.

• 5

  “Rising sea levels, tidal flooding and storm surges in southern coastal Bangladesh are causing sources of freshwater to be contaminated by saltwater, creating a critical shortage of drinking water for over 20 million people (Caretta et al., 2022; Hoque et al., 2016; SRDI, 2010). Salinity levels in most coastal aquifers and ponds in southwest Bangladesh average 1800-3000 mg/L - a level 3 to 5 times higher than the limit considered safe for human consumption (Jamei et al., 2022).” Y-RISE, “Project Description: Addressing Climate Change-Induced Drinking Water Scarcity in Coastal Bangladesh via Water Entrepreneurship,” p. 1.

• 6

  “People exposed to moderately saline (≥2000 mg/L) drinking water have a 42% greater chance of hypertension (Nahian et al., 2018), a major risk factor for cardiovascular diseases (Escober, 2002). Pregnant women drinking water with salinity above 600 mg/L are 4.4 times more likely to get preeclampsia (Khan et al., 2014), a major risk factor for maternal mortality (Goldenberg and McClure, 2011).” Y-RISE, “Project Description: Addressing Climate Change-Induced Drinking Water Scarcity in Coastal Bangladesh via Water Entrepreneurship,” p. 1.

• 7

  “A survey of well waters (n=3534) from throughout Bangladesh, excluding the Chittagong Hill Tracts, has shown that water from 27% of the 'shallow' tubewells, that is wells less than 150 m deep, exceeded the Bangladesh standard for arsenic in drinking water (50 ug L-1). 46% exceeded the WHO guideline value of 10 ug L-1 … 35 million people are believed to be exposed to an arsenic concentration in drinking water exceeding 50 ug L-1 and 57 million people exposed to a concentration exceeding 10 ug L-1.” British Geological Survey, “Arsenic contamination of groundwater in Bangladesh,” February 2021, p. i.

• 8

  “The first symptoms of long-term exposure to high levels of inorganic arsenic (for example, through drinking-water and food) are usually observed in the skin, and include pigmentation changes, skin lesions and hard patches on the palms and soles of the feet (hyperkeratosis). These occur after a minimum exposure of approximately five years and may be a precursor to skin cancer.
  In addition to skin cancer, long-term exposure to arsenic may also cause cancers of the bladder and lungs. The International Agency for Research on Cancer (IARC) has classified arsenic and arsenic compounds as carcinogenic to humans, and has also stated that arsenic in drinking-water is carcinogenic to humans.
  Other adverse health effects that may be associated with long-term ingestion of inorganic arsenic include developmental effects, diabetes, pulmonary disease, and cardiovascular disease. Arsenic-induced myocardial infarction, in particular, can be a significant cause of excess mortality.[...]
  Arsenic is also associated with adverse pregnancy outcomes and infant mortality, with impacts on child health, and exposure in utero and in early childhood has been linked to increases in mortality in young adults due to multiple cancers, lung disease, heart attacks, and kidney failure. Numerous studies have demonstrated negative impacts of arsenic exposure on cognitive development, intelligence, and memory.” WHO, “Arsenic fact sheet," 7 December, 2022.

• 9

  “We have settled on an entrepreneurship-based approach to deploying reverse osmosis (RO) plants on a large scale as the most promising strategy to address drinking water needs in this region at scale, in a way that would be financially self-sustaining.” Y-RISE, “Project Description: Addressing Climate Change-Induced Drinking Water Scarcity in Coastal Bangladesh via Water Entrepreneurship,” p. 2.

• 10

  “Reverse Osmosis (RO) technology is used to remove sodium, magnesium, lead, arsenic, and other harmful substances from water by pushing them under pressure through a semipermeable membrane.” Y-RISE, “Project Description: Addressing Climate Change-Induced Drinking Water Scarcity in Coastal Bangladesh via Water Entrepreneurship,” p. 2.

• 11

  “Desalination requires expensive reverse osmosis equipment, making it impractical for household-level implementation.” Y-RISE, “Project Description: Addressing Climate Change-Induced Drinking Water Scarcity in Coastal Bangladesh via Water Entrepreneurship,” p. 2

• 12
  • "It requires about USD 10,000 for equipment, land, and installation of a 1000L/hour capacity reverse osmosis plant that can serve the drinking and cooking water needs of 400 households for the entire year … The BRAC microfinance team … will provide business loans of up to US$10,000 to each entrepreneur to launch water businesses." Y-RISE, “Project Description: Addressing Climate Change-Induced Drinking Water Scarcity in Coastal Bangladesh via Water Entrepreneurship,” p. 2-3.
  • “BRAC will identify 1 to 2 potential water entrepreneurs through consultation, train these individuals on RO technology, provide $5,000-10,000 business set-up loans, and assist them in procuring and installing the relevant technology.” Y-RISE, “Project Description: Addressing Climate Change-Induced Drinking Water Scarcity in Coastal Bangladesh via Water Entrepreneurship,” p. 5.
  • “Community-scale RO plants will range in capacity from 500L/hour to 1500L/hour and can cover the year-round drinking and cooking water needs (4 liters per person per day) of 200 to 600 households, respectively.” Y-RISE, “Project Description: Addressing Climate Change-Induced Drinking Water Scarcity in Coastal Bangladesh via Water Entrepreneurship,” p. 5.

• 13
  • “The microfinance team has had a positive experience with >90% repayment rate and 96% business success in their small-scale pilot with a handful of entrepreneurs in the region” Y-RISE, “Project Description: Addressing Climate Change-Induced Drinking Water Scarcity in Coastal Bangladesh via Water Entrepreneurship,” p. 3.
  • “BRAC created 360 water entrepreneurs through a pilot program between 2017 to 2021. Our proposed project design is informed by the success of this pilot program.” Y-RISE, Responses to GiveWell questions of April 1, 2024 (unpublished)

• 14

  “We will implement a clustered randomized controlled trial to evaluate the effects of launching this water entrepreneurship program on profits and other business outcomes for the entrepreneurs, as well as health and socioeconomic outcomes for the ultimate beneficiaries who get access to clean water.” Y-RISE, “Project Description: Addressing Climate Change-Induced Drinking Water Scarcity in Coastal Bangladesh via Water Entrepreneurship,” p. 4.

• 15

  “We are only fundraising for evidence generation to enhance the broader impact of this approach, and not for the implementation itself.” Y-RISE, “Project Description: Addressing Climate Change-Induced Drinking Water Scarcity in Coastal Bangladesh via Water Entrepreneurship,” p. 11-12.

• 16

  “Our power calculations indicate that we would need 240 treatment communities and 125 control communities” Y-RISE, “Project Description: Addressing Climate Change-Induced Drinking Water Scarcity in Coastal Bangladesh via Water Entrepreneurship,” p. 7. (240+125=365 communities; 365 communities x 30 households = 10,950 total households)

• 17

  “[W]e will offer the loan to 1 entrepreneur in randomly divided half of the treatment communities and 2 entrepreneurs in the other half of the community. These treatment communities will be randomly assigned to two additional cross-cutting interventions. In a randomly chosen half of the treated communities, BRAC will facilitate the creation of water delivery businesses linked to each water entrepreneur. The water delivery services will take large cans of freshwater from the entrepreneurs and deliver them door-to-door to paying customers. In another independently randomly chosen half of communities, BRAC will provide discount coupons to households that can be redeemed at the water entrepreneurs’ sites. The entrepreneurs will have to be compensated by our project against the presentation of the vouchers they collect. These coupons will provide households different levels of discounts to the households enabling us to estimate the price elasticity of water demand.” Y-RISE, “Project Description: Addressing Climate Change-Induced Drinking Water Scarcity in Coastal Bangladesh via Water Entrepreneurship,” p. 6.

• 18

  Baseline surveys will occur 5-7 months after the start of the project, and endline surveys will occur 19-21 months after the start of the project (19-5=14; 21-7=14). See p. 15, Y-RISE, “Project Description: Addressing Climate Change-Induced Drinking Water Scarcity in Coastal Bangladesh via Water Entrepreneurship,” for more details on the project timeline.

• 19
  • “[W]e used hypertension as the main outcome variable of interest. This is because the medical literature has established direct links between sodium intake and hypertension through a series of randomized clinical trials (He and MacGregor, 2013; Aburto et al., 2013; Huang et al., 2020). People exposed to slightly saline (1000–2000 mg/L) and moderately saline (≥2000 mg/L) drinking water are found to have, respectively, 17% and 42% higher chance of being hypertensive compared to those drinking freshwater (Nahian et al., 2018). Y-RISE, “Project Description: Addressing Climate Change-Induced Drinking Water Scarcity in Coastal Bangladesh via Water Entrepreneurship,” p. 7.
  • “We will measure the BP of all consenting adults (age 21 and older) from each household every time … We will collect 24-hour urine samples from two consenting adults (age 21 and older) per household.” Measured via (i) blood pressure measurement and (ii) 24-hour urine sample testing.” See p. 9, Y-RISE, “Project Description: Addressing Climate Change-Induced Drinking Water Scarcity in Coastal Bangladesh via Water Entrepreneurship,” for more details on the protocols.

• 20
  • First, we will collect data on the quantity and quality of water consumed by households to determine the effect of the program on water access. The quality testing would include tests for salinity levels and microbiological contaminants. The water access module will also include data on time and money spent collecting drinking water. Second, we will collect data on the water market, including the price of water, the profits and markups of water businesses, and other business outcomes to understand the industrial organization of the water businesses that emerge. These data will allow us to more deeply understand the economic forces driving the water access results. Lastly, we collect data on health, economic, and social variables that are affected by water access. These include health outcomes such as the prevalence of hypertension, preeclampsia, stunting and wasting among children, etc. The economic indicators include labor force participation, income, consumption, etc. These outcomes will help us better understand the full welfare impacts of alleviating water scarcity. Y-RISE, “Project Description: Addressing Climate Change-Induced Drinking Water Scarcity in Coastal Bangladesh via Water Entrepreneurship,” p. 6-7.
  • Measured via (i) blood pressure measurement and (ii) 24-hour urine sample testing. See p. 9 of Y-RISE, “Project Description: Addressing Climate Change-Induced Drinking Water Scarcity in Coastal Bangladesh via Water Entrepreneurship” for more details on the protocols.
  • “We will measure the arsenic levels in water samples from each of the 360 entrepreneurs in our monthly visits. Furthermore, we will measure the arsenic levels of water samples from each of the 10,950 households in the baseline and endline surveys.” Y-RISE, Responses to GiveWell questions of April 1, 2024 (unpublished)
  • “We are also optimistic (based on our piloting results and fieldwork) that most households will buy water at the market price, and poor households will likely need only a partial subsidy. We will experimentally estimate price sensitivity to answer this question.” Y-RISE, Responses to GiveWell questions of April 1, 2024 (unpublished)
  • “We are eager to quantify the costs of illness averted … we [plan] to collect data on the occurrence of various illnesses such as diarrheal diseases.” Y-RISE, Responses to GiveWell questions of April 1, 2024 (unpublished)
  • "We will also collect data, via a survey, on reported diagnosis and treatment history of hypertension, cardiovascular diseases, preeclampsia, maternal and child morbidity and mortality, water-borne diseases, etc." Y-RISE, Responses to GiveWell questions of April 1, 2024 (unpublished)

• 21

  Y-RISE, Water Entrepreneur budget, July 2024 (unpublished). Note that after we received this budget, the cost of the grant increased from $1,243,758 to $1,299,146 (by $55,388.92), as Yale University requires Y-RISE to adopt GiveWell's Indirect Cost Policy in full (i.e. to the sub-grant as well as the main grant). We use the previous $1,243,758 budget figure to calculate cost-effectiveness estimates for this grant.

• 22

  "We already have a survey firm that we have contracted who has started collecting community level data that will allow us to target the intervention properly." Mushfiq Mobarak, email to GiveWell, March 29, 2024 (unpublished)

• 23

  Correspondence from Y-RISE to GiveWell, October 16, 2024 (unpublished)

• 24

  “For more complete context, we are currently preparing for a pilot this year. We will test out all aspects of the intervention and data collection, including bringing on board a handful of entrepreneurs, giving them microfinance loans, hiring technicians who will act as support staff, testing the water, seeing their financial performance, to try to hopefully get a well-oiled machine running by the beginning of 2025. After smoothing out any surprises and errors that come up this year, we plan to run the full scale RCT starting in early 2025. The other activity this year is to create a mapping of water salinity and water scarcity for the entire region, which will act as our sampling frame. Based on the sampling frame we plan to select the most sensible areas to launch the RCT in 2025.” Mushfiq Mobarak, email to GiveWell, July 18, 2024. (unpublished)

• 25
  • Note that a) our cost-effectiveness analyses are simplified models that are highly uncertain, and b) our cost-effectiveness threshold for directing funding to particular programs changes periodically. As of August 2024, our bar for directing funding is programs that are about 10 times as cost-effective as unconditional cash transfers. See GiveWell’s Cost-Effectiveness Analyses webpage for more information about how we use cost-effectiveness estimates in our grantmaking.
  • Note that after we received this budget, the cost of the grant increased from $1,243,758 to $1,299,146 (by $55,388.92), as Yale University requires Y-RISE to adopt GiveWell's Indirect Cost Policy in full (i.e. to the sub-grant as well as the main grant). We use the previous $1,243,758 budget figure to calculate cost-effectiveness estimates for this grant.

• 26

  “The results of the evaluation will directly inform BRAC and the Bangladesh government about the efficacy, cost-effectiveness, sustainability, and scaling potential of an entrepreneurship-based solution to a massive drinking crisis linked to a changing climate. If the results from this study look promising, BRAC has the capacity to scale up this program to reach all 20 million affected populations in coastal Bangladesh leveraging its network of branches in all the coastal subdistricts. BRAC microfinance is disbursing about two million dollars in entrepreneur loans as a part of this project, and the rigorous evidence generation through this proposed project will inform the allocation decisions for millions of dollars in future entrepreneur loans.” Y-RISE, “Project Description: Addressing Climate Change-Induced Drinking Water Scarcity in Coastal Bangladesh via Water Entrepreneurship,” p. 11.

• 27

  "With 11 million clients, BRAC MF disbursed USD 6 billion in microloans throughout Bangladesh in 2023. Therefore, if the water entrepreneurship program is found effective, BRAC can easily scale it up to reach all 20 million people affected by saltwater intrusion.” Y-RISE, Responses to GiveWell questions of April 1, 2024 (unpublished)

• 28

  “The microfinance team has had a positive experience with >90% repayment rate and 96% business success in their small-scale pilot with a handful of entrepreneurs in the region. This encouraged them to participate in this trial where they would provide loans totaling nearly USD $2 million to a couple of hundred water entrepreneurs.” Y-RISE, “Project Description: Addressing Climate Change-Induced Drinking Water Scarcity in Coastal Bangladesh via Water Entrepreneurship,” p. 3.

• 29

  Amit Kanti Sarker, Program Head for Microfinance at BRAC in conversation with GiveWell, April 25, 2024. (unpublished)

• 30

  “This new partnership will investigate … ways to bring effective, research-backed programs to scale at a whole new level. We will work as one to identify, design, evaluate, iterate, and scale programs to improve the lives of millions of people.” Yale and the World, “Yale partners with BRAC to advance the science of ending global poverty,” September 16, 2022.

• 31

  On the division of responsibilities between BRAC Microfinance and BRAC Climate Change:
  “The selection of entrepreneurs will take place in two steps:

  • BRAC Climate Change program will do the preliminary selection based on interest in doing water business, land availability, education, skills, experience, and demonstrated ability to understand the operational aspects of RO after a site visit.
  • BRAC Microfinance will do the credit screening of the individuals to make the final list of eligible entrepreneurs. BRAC MF field team collects and verifies proper documentation (National ID, duly filled application, 2 guarantors) from the interested entrepreneurs. They also verify how long that person has been residing there, the purpose of the loan, and most importantly applicant’s character. For loans above USD. 6,000, BRAC Microfinance ensures that the applicant is visited as per the procedure and approved by the second-tier managers of BRAC field offices.
  • Since Brac microfinance is lending their own money, they have strong incentives to select high-quality entrepreneurs carefully. In fact, a concern may be: “given their stringent qualification criteria for loans, would we be able to find a qualified entrepreneur in every community where we seek to work?” The pilot results described below suggest that this should not be a problem, but this is something we will need to keep an eye on.” Y-RISE, Responses to GiveWell questions of April 1, 2024 (unpublished)

• 32

  “Our research design includes a complementary effort to set up a water delivery business. With this, we seek to explore whether delivering water door-to-door instead of having community members come to the RO plant to collect increases demand.
  The research design also includes a provision for sliding-scale pricing. We will identify poor households through a listing exercise conducted at baseline, and then ask the RO entrepreneur to provide subsidized water to households identified as poor. This could be implemented either through (a) a voucher redemption system, where we distribute vouchers to poor households, and later reimburse the entrepreneur based on redeemed vouchers, or (b) by asking the entrepreneur to provide subsidized access to water in exchange for all of the ongoing training and maintenance support that BRAC will be providing them.
  In devising these plans, our goal was to subsidize water access for poor households, but the same voucher redemption procedure could be applied to target other household characteristics, such as families with a pregnant woman, or those at greater risk of hypertension. Your question is making us think that the voucher system may be the better way to go, instead of asking the entrepreneur to keep track of specific households to provide subsidized water to. And BRAC’s research/implementation team can allocate vouchers based on our own targeting criteria, which could include either poverty or health status or both.” Y-RISE, Responses to GiveWell questions of April 1, 2024 (unpublished)

• 33

  See above for a list of exploratory outcomes measured by the RCT.

• 34

  This is based on our conversation with Amit Kanti Sarker, Program Head for Microfinance at BRAC, on April 25, 2024. (unpublished)

• 35

  “Desalination requires expensive reverse osmosis equipment, making it impractical for household-level implementation.” Y-RISE, “Project Description: Addressing Climate Change-Induced Drinking Water Scarcity in Coastal Bangladesh via Water Entrepreneurship,” p. 2

• 36

  “One potential risk with RO plants is the negative externalities from brine that are generated as a waste product from the desalination process. This discharge can potentially increase the salinity level of the water bodies, aquifers, and soil areas regardless of where it is released. One potential solution could be to figure out alternative household or commercial use of brine so that it is not discharged into the local environment. For example, brine could be used in processing animal hides in the tannery industry and cooling electricity-generating equipment in thermoelectric power plants. Our implementation protocol will include a careful plan for the collection and safe discharge of brine.” Y-RISE, “Project Description: Addressing Climate Change-Induced Drinking Water Scarcity in Coastal Bangladesh via Water Entrepreneurship,” p. 14-15.

• 37

  “Thus, increased energy consumption and its negative impact on the environment appear to be the main barriers to expanding desalination technologies.” Alaward et al. 2023