Urbanisation Meets the Environment: Can Community-level Technical Assistance Induce Local Adaptation and Protect Against Flooding?

Last registered on January 10, 2025

Pre-Trial

Trial Information

General Information

Title
Urbanisation Meets the Environment: Can Community-level Technical Assistance Induce Local Adaptation and Protect Against Flooding?
RCT ID
AEARCTR-0015095
Initial registration date
December 30, 2024

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published
January 02, 2025, 7:29 AM EST

First published corresponds to when the trial was first made public on the Registry after being reviewed.

Last updated
January 10, 2025, 10:11 AM EST

Last updated is the most recent time when changes to the trial's registration were published.

Locations

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

Request Information

Primary Investigator

Affiliation
Nova School of Business and Economics

Other Primary Investigator(s)

PI Affiliation
University College London
PI Affiliation
University of Sheffield
PI Affiliation
Nova School of Business and Economics

Additional Trial Information

Status
On going
Start date
2023-11-01
End date
2025-12-31
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
Abstract
The effects of climate-related disasters such as floods are exacerbated in cities due to rapid urbanization and increasing population density, particularly in environments characterized by inadequate city planning and widespread poverty. To help communities manage risks and build resilience, we designed an intervention that provides technical knowledge and small grants to address information and liquidity constraints. Central to this intervention is the organization of community meetings to promote cooperation, mobilize local communities, and facilitate decision-making. This community-driven approach to disaster risk management empowers treated communities to select and implement their own climate adaptation projects. To assess the effects of our intervention in the coastal city of Quelimane, Mozambique, we are conducting a clustered randomized controlled trial. Beyond evaluating the intervention’s overall effectiveness, we will examine underlying mechanisms and test hypotheses regarding its effects on immediate outcomes (for example, risk awareness and community mobilization), intermediate outcomes (for example, flood mitigation behaviours among community members and leaders), and final outcomes (for example, flooding and its impact on livelihoods). To measure these, we conduct multiple rounds of surveys with 2,000 households and 500 community leaders, supplemented by data from structured community activities, systematic photographs of the drainage system, and enumerator observations, in order to broaden our pool of objective information.
External Link(s)

Registration Citation

Citation
Leeffers, Stefan et al. 2025. "Urbanisation Meets the Environment: Can Community-level Technical Assistance Induce Local Adaptation and Protect Against Flooding?." AEA RCT Registry. January 10. https://doi.org/10.1257/rct.15095-1.1
Sponsors & Partners

Sponsors

Experimental Details

Interventions

Intervention(s)
To address three critical constraints to climate adaptation infrastructure—technical knowledge, liquidity, and cooperation—we have designed a multifaceted intervention. Based on recommendations from an engineering team and tailored to the unique characteristics of each city block, we propose locally viable flood mitigation projects to overcome technical knowledge gaps. Community meetings are organized to facilitate collaborative decision-making and strengthen local cooperation. Liquidity constraints are mitigated by providing a small grant. This innovative, community-driven approach to disaster risk management empowers communities to select and implement their own climate adaptation projects.

1. Technical Knowledge
To address constraints in technical knowledge, we propose one flood mitigation project for each city block. To ensure these projects are technically sound and contextually feasible, we establish partnerships with international and local experts.
Together with the Portuguese university Instituto Superior Técnico and the Swiss engineering company MHYD, we collect relevant in-situ data, conduct hydraulic and hydrological studies, and develop a macro water model for Quelimane. In collaboration, we map the drainage canal system, identify key flood risk areas and zones suitable for mangrove replanting, and produce detailed maps to guide our intervention. The international engineering team also recommends a set of flood prevention measures, which serve as the foundation for the proposed flood mitigation projects. Additionally, we partner with local engineers to integrate their insights on existing capacities. They validate the proposed flood prevention measures for both financial and practical feasibility. To further build expertise in mangrove replanting, we partner with the Centro de Pesquisa e Tecnologia do Mar (CePTMar) and two mangrove restoration associations based in Quelimane, ANAMA and ASSOPEZA. These collaborations—both international and local—are critical to ensuring our proposed projects are technically robust and contextually appropriate.

2. Liquidity
To address liquidity constraints, we provide each city block with a small grant to support project implementation. Valued at 5,000 MZN (approximately 80 USD), the grant is announced during the first community meeting and allocated exclusively for purchasing materials needed for the project, rather than being distributed directly to the community.

3. Cooperation
To foster cooperation, we organize three rounds of community meetings. Block leaders are encouraged to invite all residents within their block to participate. To ensure adequate attendance, our team directly invites a share of residents through door-to-door mobilization activities and follows up with text message reminders.

3.1 First meeting
During the first community meeting, the community discusses flooding-related challenges. Residents identify key issues affecting their blocks and propose potential solutions. These challenges and solutions are then classified into smaller, locally manageable ones and larger issues requiring broader responses. Once the list of locally manageable projects is finalized, community members rank them by priority, and our team commits to conducting a technical evaluation before the second meeting.
Additionally, local committees, composed of the block leader and elected members, are formed during the first meeting to facilitate community decision-making. We also announce the 5,000 MZN grant per block, explaining its purpose and allocation process.

3.2 Analysis between meetings
Building on the accumulated expertise outlined earlier, we evaluate the measures proposed during the first meeting from a technical perspective. This process results in five types of flood mitigation projects, all technically sound, financially feasible, and suggested by the communities themselves: 1) cleaning drainage canals; 2) opening and connecting canals to the drainage system; 3) Entulho (using debris to fill holes where water accumulates, reducing water build-up and improving road accessibility) or Abaulamento (using debris to create an elevation in the centre of the road to direct rainwater off); 4) replanting mangroves; and 5) constructing or repairing bridges.
To identify the most suitable recommendation for each block, we develop an algorithm that incorporates: a) the community’s suggestions from the first meeting, b) the geographic location and specific characteristics of each block, and c) our engineering expertise, including the validated flood prevention measures, a detailed drainage system map, and a map of areas proposed for mangrove replanting.
This approach results in one flood-prevention project recommendation per city block, ensures that the selected project aligns with both local needs and engineering best practices, and fosters participation and ownership among community members.

3.3 Second meeting
During the second community meeting, our team presents the flood-prevention project to each block as a recommendation rather than a directive, always allowing the community to make the final choice. This approach respects the community’s deep understanding of local challenges and reinforces a sense of ownership over the selected project. To guide the discussion and support informed decision-making, our team provides detailed maps of the drainage system and mangrove replanting areas, down to the individual block level. As a result, the final project for each block is chosen through a collaborative process that combines community suggestions, our technical assessment, and a final group discussion during the second meeting.
Once a project is agreed upon, the focus shifts to identifying the materials and volunteers needed. This is still part of the second meeting. Each block receives a grant of 5,000 MZN, allocated specifically for purchasing materials required for project implementation. Our team also encourages community members to volunteer their labour and contribute materials they already possess to support the initiative, reinforcing collective effort and engagement.

3.4 Third meeting
During the third and final meeting, the community, guided by its local committee and supported by our team, comes together to implement the project using the volunteers and materials agreed-upon during the second meeting.
In addition to the primary project selected by each community, an awareness-raising campaign is implemented to promote positive flood prevention behaviours and discourage harmful practices. Many blocks suggested this activity during the first meeting, and it is a cost-effective addition to the intervention.
Focus group discussions with a block leader, a member of ASSOPEZA, and other stakeholders determine the content of the awareness-raising campaign, which comprises two phases: 1) Megaphone message: our team circulates with a megaphone, delivering concise messages in both Portuguese and the local language Chuwabu, ensuring broad understanding; 2) Community lecture: our team gathers residents for a 10-15 minute lecture, reinforcing key messages and fostering community engagement.
At the end of the lecture, communities are provided with two types of maps to enhance awareness and understanding: 1) a map of the drainage system, which identifies and categorizes the drainage canals within the block and surrounding areas, and 2) a flood risk map, which highlights the flood risk level for each block. In areas where mangrove planting is feasible, we also provide a one-page instructional guide with practical, locally tailored guidance to support successful planting efforts. This guide was developed in collaboration with CePTMar, ANAMA, and ASSOPEZA.
Finally, we sent text messages to all contacts collected during door-to-door mobilization activities and the three community meetings to reinforce the information shared in the awareness-raising campaign.
Intervention Start Date
2024-05-01
Intervention End Date
2024-12-01

Primary Outcomes

Primary Outcomes (end points)
Hypothesis 1. The intervention increases community members’ demand for leader accountability.
Hypothesis 2. The intervention increases knowledge and awareness among community members and leaders.
Hypothesis 3. The intervention increases community mobilization and cohesion.
Hypothesis 4. The intervention increases locus of control among community members and leaders.
Hypothesis 5. The intervention affects community leaders’ behaviours, namely it increases leader efforts.
Hypothesis 6. The intervention increases community members’ collective preventive behaviours regarding floods.
Hypothesis 7. The intervention affects community members’ individual preventive behaviours regarding floods.
Hypothesis 8. The intervention decreases community members’ high-risk behaviours regarding floods.
Hypothesis 9. The intervention mitigates flooding.
Hypothesis 10. The intervention decreases the impact of flooding on community members’ livelihoods.
Primary Outcomes (explanation)
Details about the Primary Outcomes, including the exact questions that will constitute the survey-based outcome variables, can be found in the pre-analysis plan, sections 5 and 6.

Secondary Outcomes

Secondary Outcomes (end points)
Details about the Secondary Outcomes, including the exact questions that will constitute the survey-based outcome variables, can be found in the pre-analysis plan, sections 5 and 6.
Secondary Outcomes (explanation)
Details about the Secondary Outcomes, including the exact questions that will constitute the survey-based outcome variables, can be found in the pre-analysis plan, sections 5 and 6.

Experimental Design

Experimental Design
1. Intervention
Details in "Intervention (Public)".

2. Sampling
There are 557 blocks in 54 neighbourhoods in 7 administrative posts in Quelimane. Building on GIS data for the Quelimane basin area and exploratory surveys with local leaders, we select 489 urban and peri-urban blocks in the city of Quelimane. This implies that 68 blocks are excluded from the sample, namely the blocks within Administrative Post 5 (due to its rurality and low population density) and the down town blocks (due to the developed drainage system in this area).

2.1 Households
To sample households, six random houses within each city block were pre-selected and ranked in QGIS. The random locations were divided into male and female respondents, to enable a gender-balanced household sample. Enumerators were instructed to visit two locations per gender and interview the household head (or partner) of the assigned gender. Interviews with unavailable respondents were rescheduled for a later date.
If any of the households were ineligible (non-residential location, non-adult household head, gender assigned not applicable), the enumerators would visit the third randomly selected location. If more substitution was needed, the enumerators were instructed to visit the closest house to the yet unused location until they had surveyed two male and two female respondents per city block.

2.2 Block Leaders
For the block leader sample, one leader per city block was surveyed. If the block leader was unavailable for the entire duration of the baseline or if the block had no assigned leader at the time, the enumerator asked the neighbourhood secretary to name the second-best person to answer our survey. This person was requested to have a block leadership position or to be a long-time resident.

3. Clustering
We employ a clustered design with a single treatment group. To achieve this, the blocks are categorized into eleven drainage zones based on watersheds and drainage systems. Within each zone, a clustering algorithm is applied utilizing the heuristic approach outlined by Wei et al. (2020), implemented through the Max-P Regionalization Algorithm from the PySAL Library.
This clustering process involves setting a minimum of three adjacent blocks and considering attribute variables such as altitude and type of drainage canal. Subsequently, we refine these clusters in a post-processing phase, prioritizing allocations consisting of three blocks and promoting increased border sharing and drainage canal connectivity. Approximately 16% of the blocks are adjusted during this phase, resulting in the definition of 159 distinct clusters.

4. Randomization
Details in "Randomization Method".

5. Statistical Power
Details in "Power calculation: Minimum Detectable Effect Size for Main Outcomes".
Experimental Design Details
Not available
Randomization Method
The 159 clusters are randomly allocated to treatment stratified by drainage zone and the number of blocks with drainage canals. This results in 80 clusters allocated to treatment, comprising 246 blocks. Note that randomization was performed at the cluster level to make sense of flood-prevention projects with clear geographical spillovers (e.g., block-level improvements in the drainage system), while treatment will be administered at the block level to leverage community cohesion and trust.
Randomization Unit
Randomization at cluster level. For details about these clusters, see 3. Clustering in "Experimental Design (Public)".
Was the treatment clustered?
Yes

Experiment Characteristics

Sample size: planned number of clusters
159 clusters. For details about these clusters, see 3. Clustering in "Experimental Design (Public)".
Sample size: planned number of observations
1956 households, 489 blocks/ block leaders, 159 clusters
Sample size (or number of clusters) by treatment arms
Treatment:
984 households
246 blocks
80 clusters

Control:
972 households
243 blocks
79 clusters

Total:
1956 households
489 blocks
159 clusters
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
We calculate the smallest effect size that can be reliably detected in a clustered randomized controlled trial using Stata's build-in power program, including 159 clusters—79 in control and 80 in treatment—for outcome variables at the household and block leader levels. Each cluster comprises two, three, or four city blocks and, on average, twelve households. The whole experimental sample encompasses 489 blocks, their leaders, and 1,956 households. We assume a 5% significance level and a 80% statistical power threshold, that is, a 80% probability that our test will yield a statistically significant result if the research hypothesis, in reality, is true. To calculate the smallest detectable effect size, we also measure the intra-cluster correlation coefficients for the different outcome variables based on household and leader baseline survey data. The baseline variables used here are standardized with mean zero and standard deviation one. We treat all clusters as if they had the same size (twelve households). The results of our power calculations, which show our study's ability to detect meaningful effects, are depicted in Figures 4 and 5 in the Pre-Analysis Plan. For household-level outcomes, we show that the experiment is powered to detect treatment effects of 0.17 standard deviations for financial contributions to community-level adaptation efforts. The minimum detectable effect size is 0.22 standard deviations for the likelihood of having been impacted by flooding last year, and 0.25 standard deviations for having seen or participated in collective action. At the local leader level, we are slightly less powered as there is only one leader per block. Here, the minimum detectable effect size is less than 0.3 standard deviations for the four chosen leader outcome variables.
IRB

Institutional Review Boards (IRBs)

IRB Name
Ethics Committee of Nova School of Business and Economics
IRB Approval Date
2023-03-13
IRB Approval Number
202255
Analysis Plan

Analysis Plan Documents

UrbanEnvMoz Pre-analysis plan

MD5: e68acf42875a5bca4e15f6a1d78d69fb

SHA1: 72fb01bc77e4b6c5c33438fddbaeaca9af49f293

Uploaded At: December 30, 2024