Community-Based Solutions to Overcome Collective Action Problems: An Experiment in Rural Vietnam.

In this project, we explore potential solutions to overcome a collective action problem within the context of a rural area in a developing country. We compare the standard approach which disseminates information about the environmental collective action problem with three novel interventions. The first equips participants with a tailored set of tools aimed at enhancing cooperation within the community and thereby facilitating collective action. The second one also provides communities with means to monitor the community in relation to the environmental problem. In a final variant the community can assess their own behaviour. We also measure the impact on spillover farmers.

In this project we explore novel solutions to overcoming environmental collective action problems in the context of a developing country. To study this, we conduct a control and three treatments plus a spillover control. Participants are randomized to one of the following:

• Baseline/Control:
Farmers in this variant will be shown a video that contains information on the environmental issue such as details on why it occurs, its costs and what can be done to mitigate it. The aim of this intervention is to replicate the most common solution to environmental issues in this context.

• Cooperation tools:
This treatment will take place with farmers in the community. Farmers will view a video that contains the same information as the control version. In addition, in this treatment the video will also provide practical tools and strategies to enhance community cooperation in addressing the issue. Further, after the video the community will have the opportunity to discuss the environmental problem together and potential solutions collectively in a community meeting. We will also create a mobile chat group for ongoing communication and help the community set community goals and a community plan to solve the environmental problem.

• Cooperation Plus Public Monitoring:
In this treatment farmers will receive the same intervention as the control and the cooperation tools treatment. In addition, farmers will also be informed that public water ways and canals will be tested for water pollution and in particular for shrimp disease and disease carriers (e.g., sediments). There will be multiple tests over the shrimp life cycle in multiple locations across the village. Farmers will not be informed of the locations before testing or the date of testing. All treated farmers in the village will be informed of the public testing results and the test location, helping them identify if others in their community have been polluting the water. The tests will be conducted by aquaculture experts.

• Cooperation Plus Public Monitoring Plus Private Testing
We combine the control and the first two interventions with an additional intervention. Farmers will also be informed of the prevalence of disease in their own shrimp ponds. In particular, we will conduct a number of tests for shrimp diseases in farmers private ponds. Farmers will have their own private ponds tested 4 times throughout the shrimp cycle. Farmers will be informed of their private pond test results, but this information will not be shared with others. The rationale for this information is that farmers may not be aware that they are polluting rivers and canals. The tests will be conducted by aquaculture experts.

• Spillover control
This sample is comprised of those who live in the treated or control villages but who did not participate in one of the above treatments. This allows us to measure treatment spillovers.

The experimental design can be summarized as follows:
Control: environ info only
Treatment 1: environ info + info on cooperation
Treatment 2: environ info + cooperation+ public monitoring
Treatment 3: environ info + cooperation+ public monitoring + private testing
Treatment 4: Spillover control

2024-10-02

2024-12-20

We will collect both short- and medium-term outcomes.
Short term outcomes: These are measured directly after the intervention

-Perceptions about the Community: We will use question s5_q2 to measure this. Further, we also use an index taken from a set of questions examining attitudes about the community. The index will be created by taking the mean of all the individual components. In both cases we will compare the cooperation treatments relative to the non-cooperation treatment.

Medium term outcomes: These are measured at least 3 months after the intervention (the exact time depends on availability)

-Shrimp farming production: We will collect a measure of total shrimp production in the relevant shrimp crop.

-Prevalence of the disease: We will elicit the proportion of shrimp impacted by disease as a total proportion of their current shrimp production.

- How much the farmer spent on treating the water released into public canals and rivers

See above

To understand possible mechanisms, we elicit the following secondary outcome variables:
-Contribution to the public good. This is measured using a standard public goods game.

-A dummy variable equal to one if the individual is doing something to overcome the environmental problem.

- A measure of the frequency with which an individual has discussed the environmental problem outside the households.

-Measure of social capital- this is an index created from the set of survey questions measuring social capital. The index is created by taking the average of all the survey questions. The index is measured at the household level.

-The amount spent on treating water taken in from public rivers and canals

See above

In this project, we explore potential solutions to address a collective action problem in a rural area of a developing country. We compare the standard approach of disseminating information about the environmental issue with three innovative interventions. The first provides participants with tailored tools designed to strengthen community cooperation and promote collective action. The second adds a mechanism for communities to monitor their peers' behavior in relation to the environmental problem. In the final variant, the community members are given information to assess their own behavior. We also measure the impact on spillover farmers.

The experiment is conducted in Vietnam. The environmental problem is water pollution, and the spread of shrimp disease caused by the dumping of untreated and diseased water into public water ways. This water pollution harms the quality of the water, reducing the production of shrimp farmers in the local area.

In this section we outline further details on the experimental design.

Setting:
Based on focus group discussion in the communities of the Mekong delta it is clear that a key environmental problem is water pollution especially as a result of discharge from shrimp farming. Many shrimp farmers discharge polluted water which contain diseases from their shrimp pools directly harming other local farmers. This issue resembles a standard collective action problem, wherein the individual incentive for farmers to act is low, particularly since it's unclear who discharged the water.

Experimental Details:
The experiment will be conducted in 62 shrimp farming villages. Within each village the intervention will be conducted with 24 farmers plus 10 spillover farmers. To participate farmers must have an active shrimp farm and be responsible for making decisions on the shrimp farm. Farmers will be assigned to one of the following interventions.

Baseline Environmental Info Only (control):
Farmers in this variant will be shown a video. In this video we include information on the environmental issue such as details on why it occurs, its costs and what can be done to mitigate it (excluding solutions that require the involvement of others).

T1: Environmental Info + Cooperation tools
This treatment will take place with farmers in the community. Farmers will view a video that contains the same information as the control version. In addition, in this treatment the video will provide practical tools and strategies to enhance community cooperation in addressing the issue. Further, after the video the community will have the opportunity to discuss the environmental problem together and potential solutions collectively in a community meeting. We will also create a mobile chat group for ongoing communication and to help the community set community goals and a community plan to solve the environmental problem.

T2: Environmental Info + Cooperation tools+ Public Monitoring:
This treatment contains T1 and T2. In addition, farmers will also be informed that public water ways and canals will be tested for water pollution and in particular for shrimp disease and disease carriers (e.g., within sediments). This information will be distributed at the conclusion of the community discussion. Public canals will be tested at four different times. At most 6 public locations will be tested within a village. Community members will be informed that testing of public canals will occur, but they will not be notified of the specific locations, times, or frequency of the tests. Results will be made public to the community via the mobile chat group and also individually via SMS.

T3: Environmental Info + Cooperation tools + Public Monitoring AND Private Testing:
This treatment includes the baseline environmental info, T1 and T2. We also add private testing. Farmers will also be able to select a pond which can be tested for two common diseases. We will conduct four tests across the shrimp cycle per household. Farmers will be informed of their private pond test results, but this information will not be shared with others. The tests will be conducted by aquaculture experts.

T4: Spillover control
This sample is comprised of those who live in the treated or control villages but who did not participate in one of the above treatments. This allows us to measure treatment spillovers

Below we outline how we will use our primary outcomes and the key hypothesis.

We generate the following hypothesis:
Our hypothesis below focuses on our medium term outcomes.

Hypothesis 1: Providing cooperation tools will increase shrimp production, the amount spent on water treatment and decrease disease prevalence relative to the control.

Rationale:
• Greater cooperation and community discussion about the disease will allow farmers to communicate and discuss how to collectively solve the environmental problem improving shrimp production, the amount spent on treatment and reduce prevalence in treated farms. This will also reduce disease prevalence and increase shrimp production in spillover farms.

• The cooperation treatment will give farmers a community plan and goals to overcome the environmental issue. This will raise outcomes in treated and spillover farms in the village.

• It will also increase the social cost of polluting public water, also improving outcomes in treated and spillover farms in the village.

Hypothesis 2: Providing disease test results from public spaces will increase shrimp production, the amount spent on water treatment and decrease disease prevalence relative to the cooperation tools treatment (T1).

Rationale:
• Social sanctions: Testing of public canals and water way may increase the ability to social sanction (e.g. social exclusion, losing trust/reputation, etc.) those who dump diseased water. Aware of this, farmers may reduce their dumping practices, leading to increased production and lower disease prevalence in both treated and spillover farms. There are two potential mechanisms at play: 1) Farmers may reduce harmful behavior in anticipation of social sanctions; 2) Actual social sanctions may be imposed, further discouraging harmful practices.

• The effectiveness of public monitoring will also be dependent on the social capital in the village. The cooperation tools will create a platform to discuss monitoring information and disease control measures.

• Private benefit: Public testing will inform farmers of the disease prevalence in local water ways. When local waterway disease prevalence is high, farmers have a greater incentive to treat water taken in. Without this knowledge farmers may be unaware of the disease prevalence and may be less likely to treat incoming water. This will increase shrimp production, the amount spent on water treatment and decrease disease prevalence.

Hypothesis 3: Providing private test results will increase shrimp production, the amount spent on water treatment and decrease disease prevalence relative to cooperation + public treatment (T2).

Rationale:
Private benefit: Private disease testing may increase the detection of disease in private ponds and give farmers time to mitigate the spread of disease across ponds. This will increase treated farmers production and decrease disease prevalence. This will also improve outcomes in spillover farms.

Alternatively, as farmers are more able to detect disease within their own ponds, they may increase dumping to prevent the spread of disease within their ponds. While this could boost production and reduce disease prevalence for treated farmers, if many farmers follow this practice, it could lead to higher disease prevalence and reduced production across both treated and spillover farms. The impact will also depend on how much effort and resources farmers invest on treating the diseased water before releasing it into the canals and the potential social cost of getting caught (from the public testing).

• Ignorance: When farmers are informed of their private test results, they become aware of the prevalence of disease within their ponds, this means that the cost of dumping is far more explicit. In other words, farmers can’t hide behind ignorance of not knowing ponds are diseased when dumping water. This should reduce dumping of diseased water, increase spending on treatment and reduce the spread of disease, improving outcomes in treated and untreated farms.

At the aggregate level we expect the public+private positive benefits to have a larger impact over the negative effects of the private information.

In this study, a sampling approach will be employed to increase the representativeness of the sample. The sampling process will involve two stages: village selection and household selection.

First, four districts that are key shrimp producers are selected. A list of all villages in these four districts is obtained from the local government. The number of shrimp farms in each village is also obtained. From this village census, we exclude all villages that have less than 40 shrimp households. We then randomly select 74 villages and use the first 62 for our sample and the next 12 as possible replacements. We use stata to generate a random number for each village. A treatment is then assigned based on the ordering of the random number.

Second, in each selected village, we will obtain the list of all shrimp farmers from the local authorities and randomly select 24 farmers to participate. In cases when a household list cannot be obtained, we will work with the local village chief to ensure that the sample is close to being representative of shrimp farmers in the village.

The cluster unit will be the village level. This is because the treatments are assigned at the village level.

Yes

We will randomly select 62 villages==62 clusters.

Given the budget constraints and required logistics, we expect to reach about 2108 observations across 62 villages.

The sample will be distributed across the treatments as follows:

Baseline Environmental Info Only: 20 villages. With 24 observations in each village, for a total sample of 480 observations.

T1-T3: 14 villages in each treatment with 24 observations in each village, for a total sample of 336 observations in each treatment (T1 to T3).

Spillover control: The spillover control will be conducted in all villages. We will collect 10 observations in each village, for a total sample of 620 observations.

Under some strict assumptions we calculate the MDE of disease prevalence. Assuming a control disease prevalence rate of 25% and focusing on pairwise comparisons we will have an MDE of 0.084. Using similar assumptions our MDE for shrimp production is about 0.30 tons.