|
Field
Trial Status
|
Before
in_development
|
After
on_going
|
|
Field
Abstract
|
Before
At least 2 billion people around the world drink water from a contaminated source. About two million children die every year from diarrheal diseases and around 700 million people worldwide are affected by chronic kidney disease because of unsafe drinking water and sanitation, despite the availability of effective and inexpensive technologies to improve water quality (Null et al. 2012). Most efforts to provide clean water focused on the technology of chlorine (adding chlorine or chlorine compounds such as sodium hypochlorite to water), which is affordable and easy to use (see, for example, Amrita, Kremer, and Zwane 2010; Null et al. 2012; Dupas et al. 2022).
Despite the evidence of health benefits associated with chlorinated water, empirical evidence indicates that most households in developing countries are not willing to adopt and pay much for it (Amrita, Kremer, and Zwane 2010; Null et al. 2012; Berry, Fischer, and Guiteras 2020). Indeed, Dupas et al. (2016) and Dupas et al. (2022) report that less than half of households consume chlorinated water even when they receive chlorine for free.
A plausible hypothesis is that previous efforts to promote adoption of clean water disregard the importance of understanding the local culture, namely local taste and tradition. Our hypothesis is that dislike for taste of chlorinated water and/or its cultural unsuitability might explain the low take-up rates (and willingness to pay) of chlorinated water, other than price itself or lack of adequate information. Moreover, in rural areas with high illiteracy rates, people may use taste as a strong signal of healthy water more than standard information messages.
Specifically, we provide experimental evidence that local taste and traditions are important –yet overlooked so far– dimensions in the decision of adopting clean water. We introduce a new culturally-friendly technology that produces clean water taking into account local culture and preferences, by filtering local water in water treatment units. Differently from chlorination, this technology does not alter the taste of water, nor the taste of traditional beverages that are a relevant part of locals’ consumption habits and sociality. Moreover, in rural areas with high illiteracy rates, people may use taste as a strong signal of healthy water.
We study this question in rural Egypt, where groundwater sources exhibit increasing levels of contamination and salinity due to seawater intrusion and soil degradation (Said et al., 2021). We perform four different experiments using samples of filtered and chlorinated water. While the two types of water are healthier than status-quo water, they differ in taste, since filtered water is closer to the taste individuals expect in local water and traditional beverages. The first experiment consists of a blind test where respondents test both a sample of filtered and chlorinated water, not knowing the types of water, and then indicate their preferred sample. Second, we test if filtered water leads to higher adoption rates and willingness-to-pay for clean water using the Becker et al. (1964) mechanism. The third experiment explores underlying reasons behind the preferences revealed in the previous experiments by investgating individuals' perceptions of taste, health and cultural suitability of filtered and chlorinated water. Lastly, we inspect long-run health and socio-economic outcomes following adoption of treated and chlorinated water.
|
After
At least 2 billion people around the world drink water from a contaminated source. About two million children die every year from diarrheal diseases and around 700 million people worldwide are affected by chronic kidney disease because of unsafe drinking water and sanitation, despite the availability of effective and inexpensive technologies to improve water quality (Null et al. 2012). Most efforts to provide clean water focused on the technology of chlorine (adding chlorine or chlorine compounds such as sodium hypochlorite to water), which is affordable and easy to use (see, for example, Amrita, Kremer, and Zwane 2010; Null et al. 2012; Dupas et al. 2022).
Despite the evidence of health benefits associated with chlorinated water, empirical evidence indicates that most households in developing countries are not willing to adopt and pay much for it (Amrita, Kremer, and Zwane 2010; Null et al. 2012; Berry, Fischer, and Guiteras 2020). Indeed, Dupas et al. (2016) and Dupas et al. (2022) report that less than half of households consume chlorinated water even when they receive chlorine for free.
A plausible hypothesis is that previous efforts to promote adoption of clean water disregard the importance of understanding the local culture, namely local taste and tradition. Our hypothesis is that dislike for taste of chlorinated water and/or its cultural unsuitability might explain the low take-up rates (and willingness to pay) of chlorinated water, other than price itself or lack of adequate information. Moreover, in rural areas with high illiteracy rates, people may use taste as a strong signal of healthy water more than standard information messages.
Specifically, we provide experimental evidence that local taste and traditions are important –yet overlooked so far– dimensions in the decision of adopting clean water. We introduce a new culturally-friendly technology that produces clean water taking into account local culture and preferences, by filtering local water in water treatment units. Differently from chlorination, this technology does not alter the taste of water, nor the taste of traditional beverages that are a relevant part of locals’ consumption habits and sociality. Moreover, in rural areas with high illiteracy rates, people may use taste as a strong signal of healthy water.
We study this question in rural Egypt, where groundwater sources exhibit increasing levels of contamination and salinity due to seawater intrusion and soil degradation (Said et al., 2021). We perform several experiments using samples of filtered and chlorinated water. While the two types of water are healthier than status-quo water, they differ in taste, since filtered water is closer to the taste individuals expect in local water and traditional beverages. The first experiment consists of a blind test where respondents test both a sample of filtered and chlorinated water, not knowing the types of water, and then indicate their preferred sample. Second, we test if filtered water leads to higher adoption rates and willingness-to-pay for clean water using the Becker et al. (1964) mechanism. Three ancillary experiments explore underlying reasons behind the preferences revealed in the previous experiments by investigating individuals' perceptions of taste, health and cultural suitability among samples of filtered, chlorinated and bottled water. Lastly, we inspect long-run health and socio-economic outcomes following adoption of treated and chlorinated water.
|
|
Field
Trial End Date
|
Before
December 31, 2023
|
After
December 31, 2025
|
|
Field
Last Published
|
Before
June 19, 2023 05:01 AM
|
After
April 27, 2024 11:09 AM
|
|
Field
Intervention (Public)
|
Before
First Experiment: Blind Test
The first experiment explores whether individuals prefer chlorinated water or filtered water. This experiment consists of a “blind test” between samples of chlorinated water and filtered water. Eligible population includes men and women aged 18 or more who are permanent residents of the villages. Each individual tests both a samples of filtered and chlorinated water, not knowing the types of water, and then indicates the preferred sample. Since, in Egypt, tea is an everyday beverage drank during meals, shared in social gatherings, and as an alternative to water during the day, the taste of tea made with clean water could affect people’s decision to adopt. We explore for this by randomly assigning half of the individuals in the blind test to taste plain water (both filtered and chlorinated) and the other half to taste tea (both made with filtered and chlorinated water). To avoid potential biases, we randomly assign the order in which individuals taste the samples.
Second Experiment: Willingess-to-Pay
In the second experiment we measure willingness to pay. This experiment focuses on women, who are typically the ones in charge of fetching water. We randomly assign half women to the treatment group (filtered water) and half to the control group (chlorinated water). After trying a sample of the water, the woman states her willingness to pay for a 20-liter bucket of water. To obtain women’s willingness-to-pay, the woman first states an offer price for the 20 liters of water. Then a random transaction price is drawn (in our context choosing one envelope from an unmarked set). If the random transaction price is greater than the offer price, the woman cannot purchase the product. If the random price is less than or equal to the offer price, she has to purchase the product, and pays the random transaction price draw rather than the stated offer. For expected utility maximizers, the optimal strategy is to bid the true maximum willingness to pay (Becker, Degroot, and Marschak 1964).
Third experiment: Mechanisms
The third experiment explores underlying reasons behind the preferences revealed in the first experiment. However, filtered water is both healthier and tastier than chlorinated water, so the reasons for filtered water being preferred are not yet known. For this third experiment, we produce a new water that is as healthy as filtered water, but that taste as the local chlorinated water. That is, the two waters are equally healthy but differ in taste. Individuals are randomly assigned to taste either filtered (treatment) or chlorinated water (control) and, after tasting, are asked to describe the water in terms of its taste, health and cultural suitability (i.e. suitability to prepare traditional food and beverages). This allows us to identify the dimension(s) in which filtered water is superior to chlorinated water. On top of that, before tasting the water, we inform all participants that the water they will taste is healthy. In this way, we equalize actual information on health. Therefore, remaining differences in health perceptions can be attributed only to taste, which is the only signal about the quality of water that the individual receives after the disclosure of information.
Fourth Experiment: Health
In the fourth experiment, we randomly assign households to 3 groups: a treatment group, a control group, and a pure control group. The treatment group will receive a free voucher to obtain filtered water delivered at home for 2 months. The control group will receive a free voucher to obtain chlorinated water delivered at home for 2 months. Finally, the pure control group will not be offered any clean water. A follow-up visit will take place after 2 months, and we will survey household both in the treated and in the control group in order to record medium-term outcomes. In the medium term, there is a direct link between consumption of dirty water and diseases such as diarrhea and kidney problems. In addition, fetching clean water is time-and-labor-intensive, which could detract from education or productive activities, a burden that falls disproportionately on women and children. Lastly, since women are responsible for domestic water management (e.g., cleaning dishes, doing laundry, preparing food), they are usually blamed by their husbands for water-related issues, increasing the chances of intimate partner violence. That is, we will study the medium-term effect of clean-water adoption on water-related diseases, women’s working decisions, children’s education, and intimate partner violence.
|
After
First Experiment: Blind Test
The first experiment explores whether individuals prefer chlorinated water or filtered water. This experiment consists of a “blind test” between samples of chlorinated water and filtered water. Eligible population includes men and women aged 18 or more who are permanent residents of the villages. Each individual tests both a samples of filtered and chlorinated water, not knowing the types of water, and then indicates the preferred sample. Since, in Egypt, tea is an everyday beverage drank during meals, shared in social gatherings, and as an alternative to water during the day, the taste of tea made with clean water could affect people’s decision to adopt. We explore for this by randomly assigning half of the individuals in the blind test to taste plain water (both filtered and chlorinated) and the other half to taste tea (both made with filtered and chlorinated water). To avoid potential biases, we randomly assign the order in which individuals taste the samples.
Second Experiment: Willingness-to-Pay
In the second experiment we measure willingness to pay. This experiment focuses on women, who are typically the ones in charge of fetching water. We randomly assign half women to the treatment group (filtered water) and half to the control group (chlorinated water). After trying a sample of the water, the woman states her willingness to pay for a 20-liter bucket of water. To obtain women’s willingness-to-pay, the woman first states an offer price for the 20 liters of water. Then a random transaction price is drawn (in our context choosing one envelope from an unmarked set). If the random transaction price is greater than the offer price, the woman cannot purchase the product. If the random price is less than or equal to the offer price, she has to purchase the product, and pays the random transaction price draw rather than the stated offer. For expected utility maximizers, the optimal strategy is to bid the true maximum willingness to pay (Becker, Degroot, and Marschak 1964).
Third experiment: Mechanisms and ancillary
The third experiment explores underlying reasons behind the preferences revealed in the first experiment. However, filtered water is both healthier and tastier than chlorinated water, so the reasons for filtered water being preferred are not yet known. For this third experiment, we produce a new water that is as healthy as filtered water, but that taste as the local chlorinated water. That is, the two waters are equally healthy but differ in taste. Individuals are randomly assigned to taste either filtered (treatment) or chlorinated water (control) and, after tasting, are asked to describe the water in terms of its taste, health and cultural suitability (i.e. suitability to prepare traditional food and beverages). This allows us to identify the dimension(s) in which filtered water is superior to chlorinated water. On top of that, before tasting the water, we inform all participants that the water they will taste is healthy. In this way, we equalize actual information on health. Therefore, remaining differences in health perceptions can be attributed only to taste, which is the only signal about the quality of water that the individual receives after the disclosure of information.
Two ancillary experiments are aimed at disentangling which aspect of taste is most relevant between dislike for the taste of chlorine, or a cultural preference for local ancestral water. First, we replicate the above Mechanisms experiment in our sample of rural villages, but chlorinated water is substituted by a leading commercial water brand (E.g. Evian water), thus individuals taste either filtered water or bottled water. Since both water types are comparably healthy and devoid of chlorine taste, the experiment should discern the relative significance of chlorine taste versus local taste of water while keeping constant both water healthiness and the information provided on water healthiness. Last, to further examine the role of local ancestral taste, we replicate the experiment in Cairo with participants from various countries and parts of Egypt, but not from our rural sample, for whom thus our filtered water is not their local ancestral water, since it is tailored to resemble the taste of local water in our target rural areas but likely not the taste of water from participants' diverse geographical locations.
Fourth Experiment: Health
In the fourth experiment, we randomly assign households to 3 groups: a treatment group, a control group, and a pure control group. The treatment group will receive a free voucher to obtain filtered water delivered at home for 2 months. The control group will receive a free voucher to obtain chlorinated water delivered at home for 2 months. Finally, the pure control group will not be offered any clean water. A follow-up visit will take place after 2 months, and we will survey household both in the treated and in the control group in order to record medium-term outcomes. In the medium term, there is a direct link between consumption of dirty water and diseases such as diarrhea and kidney problems. In addition, fetching clean water is time-and-labor-intensive, which could detract from education or productive activities, a burden that falls disproportionately on women and children. Lastly, since women are responsible for domestic water management (e.g., cleaning dishes, doing laundry, preparing food), they are usually blamed by their husbands for water-related issues, increasing the chances of intimate partner violence. That is, we will study the medium-term effect of clean-water adoption on water-related diseases, women’s working decisions, children’s education, and intimate partner violence.
|
|
Field
Intervention End Date
|
Before
December 31, 2023
|
After
December 31, 2025
|
|
Field
Experimental Design (Public)
|
Before
First experiment (blind test), second experiment (willingness-to-pay) and third experiment (mechanisms): The first three experiments are conducted in the same 7 villages in rural Upper Egypt. We propose a randomized design at the individual level, stratified by village. We recruit different individuals for each experiment. In the first and third experiment, we recruit individuals in the village aged 18 or more. In the second experiment, the respondent will be the woman in charge of the household, who is typically the one that fetches water.
In the first experiment, we randomly assign half of the sample to taste both filtered and chlorinated water (n=100), and the other half to taste both filtered and chlorinated tea (n=100). Additionally, to avoid potential biases related to the order of testing the samples, within each treatment arm, half of the individuals are randomly assigned to taste chlorinated water first and the other half to taste filtered water first.
In the second experiment, half of the sample (n=200) is randomly assigned to the treatment group (taste filtered water and elicit willingess-to-pay) and half (n=200) to the control group (taste chlorinated water and elicit willingess-to-pay).
In the third experiment, half of the sample (n=100) is randomly assigned to taste filtered water (treated group) and half (n=100) chlorinated water (control group). In each of the three experiments, we will administer the treatment followed by the related survey measures.
Fourth experiment (health): It is conducted in the same 7 villages in rural Upper Egypt. We propose a randomized design at the household level, stratified by village. We randomly assign househols to 3 groups: a treatment group (n=700), a control group (n=700), and a pure control group (n=700). We conduct a survey after 2 months from the intervention.
|
After
First experiment (blind test), second experiment (willingness-to-pay) and third experiment (mechanisms+2 ancillary): The first three experiments and the first ancillary experiment are conducted in the same 8 villages in Asyut, in rural Upper Egypt. We propose a randomized design at the individual level, stratified by village. We recruit different individuals for each experiment. In the first and third experiment, we recruit individuals in the village aged 18 or more. In the second experiment, the respondent will be the woman in charge of the household, who is typically the one that fetches water. The second ancillary experiment is conducted in Cairo, where we recruit individuals aged 18 or older with diverse geographic backgrounds who were participating to a water fair.
In the first experiment, we randomly assign half of the sample to taste both filtered and chlorinated water (n=100), and the other half to taste both filtered and chlorinated tea (n=100). Additionally, to avoid potential biases related to the order of testing the samples, within each treatment arm, half of the individuals are randomly assigned to taste chlorinated water first and the other half to taste filtered water first.
In the second experiment, half of the sample (n=200) is randomly assigned to the treatment group (taste filtered water and elicit willingess-to-pay) and half (n=200) to the control group (taste chlorinated water and elicit willingess-to-pay).
In the third experiment, half of the sample (n=100) is randomly assigned to taste filtered water (treated group) and half (n=100) chlorinated water (control group). In the first ancillary experiment, half of the sample (n=100) is randomly assigned to taste filtered water (treated group) and half (n=100) commercial water (control group) in the eight villages in Asyut. In the second ancillary experiment, half of the sample (n=100) is randomly assigned to taste filtered water (treated group) and half (n=100) commercial water (control group) in Cairo.
In each of the three experiments, we will administer the treatment followed by the related survey measures.
Fourth experiment (health): It is conducted in the same 7 villages in rural Upper Egypt. We propose a randomized design at the household level, stratified by village. We randomly assign households to 3 groups: a treatment group (n=700), a control group (n=700), and a pure control group (n=700). We conduct a survey after 2 months from the intervention.
|
|
Field
Randomization Method
|
Before
First experiment (blind test), second experiment (willingness-to-pay) and third experiment (mechanisms): randomization done on the spot with device
Fourth experiment (health): randomization done by a computer
|
After
First experiment (blind test), second experiment (willingness-to-pay) and third experiment (mechanisms+ancillary): randomization done on the spot with device
Fourth experiment (health): randomization done by a computer
|
|
Field
Randomization Unit
|
Before
First experiment (blind test), second experiment (willingness-to-pay) and third experiment (mechanisms): Random assignment will be at the individual level. Our experimental design stratifies at the village level.
Fourth experiment (health): Random assignment will be at the household level. Our experimental design stratifies at the village level.
|
After
First experiment (blind test), second experiment (willingness-to-pay) and third experiment (mechanisms+ancillary): Random assignment will be at the individual level. Our experimental design stratifies at the village level.
Fourth experiment (health): Random assignment will be at the household level. Our experimental design stratifies at the village level.
|
|
Field
Planned Number of Observations
|
Before
First experiment (blind test): 200 individuals
Second experiment (willingness-to-pay): 400 women
Third experiment (mechanisms): 200 individuals
Fourth experiment (health): 2100 households
|
After
First experiment (blind test): 200 individuals
Second experiment (willingness-to-pay): 400 women
Third experiment: 200 individuals (mechanisms), 200 individuals (1st ancillary), 200 individuals (2nd ancillary)
Fourth experiment (health): 2100 households
|
|
Field
Sample size (or number of clusters) by treatment arms
|
Before
First experiment (blind test): 100 individuals (filtered and chlorinated water), 100 individuals (filtered and chlorinated tea)
Second experiment (willingness-to-pay): 200 women (filtered water), 200 women (chlorinated water)
Third experiment (mechanisms): 100 individuals (filtered water), 100 individuals (chlorinated water)
Fourth experiment (health): 700 households (filtered water), 700 households (chlorinated water), 700 households (pure control)
|
After
First experiment (blind test): 100 individuals (filtered and chlorinated water), 100 individuals (filtered and chlorinated tea)
Second experiment (willingness-to-pay): 200 women (filtered water), 200 women (chlorinated water)
Third experiment (mechanisms): 100 individuals (filtered water), 100 individuals (chlorinated water)
1st ancillary (rural villages): 100 individuals (filtered water), 100 individuals (commercial water)
2nd ancillary (Cairo): 100 individuals (filtered water), 100 individuals (commercial water)
Fourth experiment (health): 700 households (filtered water), 700 households (chlorinated water), 700 households (pure control)
|