The water risk game: a Lab-in-the- field experiment

Last registered on August 06, 2020


Trial Information

General Information

The water risk game: a Lab-in-the- field experiment
Initial registration date
August 05, 2020
Last updated
August 06, 2020, 10:07 AM EDT



Primary Investigator

University of Oxford

Other Primary Investigator(s)

PI Affiliation
University of Oxford

Additional Trial Information

In development
Start date
End date
Secondary IDs
Achieving universal and equitable water service provision by 2030 represents a key priority in the international policy agenda. However, meeting this target seems yet out of reach.

In 2015, 2.1 billion people lacked safely managed drinking water services globally and 844 million people did not have basic drinking water services (WHO/UNICEF, 2017). Around a million handpumps in rural Africa provide water to approximately 200 million rural Africans but break frequently, wasting billions of dollars of investment (Baumann, 2009; Baumann and Furey, 2013) and forcing the poor to regularly use more distant and often dirty water sources. Recent estimates show that one in four handpumps in sub-Saharan Africa is non-functional (Foster et al., 2019) and this poses serious risks to water security.

To reduce this risk novel professional maintenance models for rural water infrastructure are spreading across Africa. Intuitively, these maintenance models ``insure'' the handpump and protect communities against pump malfunctioning and the corresponding water insecurity. Practically, subscribing to these maintenance schemes entails that communities sign an annual contract and make regular monthly payments to the service provider to be guaranteed an effective and prompt repair and thus enjoy uninterrupted clean water provision.

Given that water supply in rural Africa tends to be a community affair, water users are required to cooperate and invest part of their private money to be able to pay the premium and insure the pump against its breakage. Herein lies the dilemma: the subscription to the professional maintenance service relies on voluntary contributions from water-users, which are beneficial for the group, but this comes at a cost for single individuals who could have chosen to invest elsewhere. Moreover, failing to collect the needed amount of money to subscribe to the maintenance service is likely to result in water insecurity for the entire community.

Will water-users reach a collective agreement to ``insure'' the handpump through individual contributions when everybody would suffer severe consequences from the handpump break down if the community did fail to cooperate?

We designed what we call a "the water-risk game", a one-shot public good game with threshold. Each subject in our experiment will face the same tradeoff: the more she invests into the collective good, the higher the probability that the threshold is reached and water security will be enjoyed by all members of the community, but this comes at the expenses of her private money which she is guaranteed to receive at the end of the game if the target sum has been reached. Conversely, failing to reach the target sum implies that each individual will preserve their private money but the collective faces risks. Our intervention consists in manipulating the probability with which the collective risks water security.

The main outcome variable we aim to capture is the group-wide contribution level and thus the share of groups who fail to meet the target, resulting from different risk levels of water insecurity. We expect the percentage of groups failing to meet the threshold to decrease with the probability of losing water security. Equivalently, group contributions are supposed to increase as exposure to risk increases.
We will analyse the dynamics of this outcome across treatments and between groups who have already subscribed to a pump maintenance scheme and those who have not.

We will also register individual outcomes. Specifically, by means of their individual contributions we will be able to identify in each group altruists, fair sharers and free-riders.

Our main objective is to study whether uncertainty in water security represents a key reason why communities in Africa fail to cooperate and subscribe to new maintenance service models.
Aside from a simple treatment comparison across arms, our analysis will distinguish between insured and not insured handpumps and analyse the key reasons behind the community choice. Additionally, we will be able to gain a more nuanced understanding of the observed variation in contribution patterns thanks to data that we will collect by means of a survey instrument which we will administer upon completion of the experimental procedure.
External Link(s)

Registration Citation

Innocenti, Stefania and Johanna Koehler. 2020. "The water risk game: a Lab-in-the- field experiment." AEA RCT Registry. August 06.
Experimental Details


Our intervention consists in manipulating the probability with which the collective risks water security.
Intervention Start Date
Intervention End Date

Primary Outcomes

Primary Outcomes (end points)
The main outcome variable we aim to capture is the total group level contributions and more specifically the fractions of groups which fail to meet the threshold, resulting from different risk levels of water insecurity.
Primary Outcomes (explanation)

Secondary Outcomes

Secondary Outcomes (end points)
Given our design, individuals decisions can be recognised as altruists, fair sharers, and free riders depending on whether the participant contributes 0, 2 or 4 tokens (i.e. the entire private endowment) to meet the target T. We will thus record the number of free-riders, fair sharers or altruists in each group in each treatment arm.
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
Ours is a one-shot public good game with threshold.
Experimental Design Details
120 groups of six subjects will participate in our experiment. Each participant will receive a `personal envelope' with an endowment of 4 tokens. Additionally, there will be a `collective envelope' which will contain 24 tokens. Participants will be grouped in groups of size 6. Each group member will be asked to invest 0, 2 or 4 tokens from their personal envelope into a ``water fund''. If the total group contributions to the water fund reach or surpass the target which is set to be equal to 12 tokens, all group members will receive what they have not invested in the water fund plus the content of the `collective envelope' equally split among the 6 participants. If the group instead fails to collect the target, each person will be refunded of her investment so that her private envelope is fully preserved. Yet, the `collective envelope' will be lost with 90%, 50%, 10% probability depending on the treatment arm.
Randomization Method
Randomization done in office by a computer
Randomization Unit
The treatment will be randomised at the hand pump level.
Was the treatment clustered?

Experiment Characteristics

Sample size: planned number of clusters
120 handpumps
Sample size: planned number of observations
A total of 720 subjects will take part into the the game (6 participants for each of the 120 hand pumps)
Sample size (or number of clusters) by treatment arms
The treatment arms will be block-randomised at the pump level. Specifically,
we will select a grand-total of 120 pumps among the 300 working pumps in
Kwale. 60 pumps will be selected among those which are currently enrolled
in a maintenance scheme and 60 which yet have not subscribed to this type
of service. For each one of these 2 groups, 20 handpumps will be assigned to
the high risk treatment, 20 to the medium risk and 20 to the low risk arm.
At the pump level, we will use rosters to select 6 families among the water-
users. One individual per family will be then asked to
participate to the experiment.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)

Institutional Review Boards (IRBs)

IRB Name
Central University Research Ethics Committee (CUREC) of the University of Oxford
IRB Approval Date
IRB Approval Number
SOGE 1A2020-45


Post Trial Information

Study Withdrawal

There are documents in this trial unavailable to the public. Use the button below to request access to this information.

Request Information


Is the intervention completed?
Data Collection Complete
Data Publication

Data Publication

Is public data available?

Program Files

Program Files
Reports, Papers & Other Materials

Relevant Paper(s)

Reports & Other Materials