Incentives and Maintenance for Rainwater Harvesting Systems in Mexico City

This randomized controlled trial tests whether improving maintenance and providing incentives enhance the effective use of rainwater harvesting (RWH) systems among urban households in Mexico City. A total of 750 households with functional RWH systems are randomized in two stages at the household level, stratified by location and baseline storage capacity. In Stage 1 (~2 months), households are assigned to one of four groups: (i) Information + messages, (ii) Hands-on training, (iii) Regular maintenance, or (iv) Control. In Stage 2 (~2 months), within each Stage 1 arm, 50% of households are also assigned to a per-liter cash incentive tied to rainwater capture. Primary outcomes include rainwater volumes captured (metered) and water obtained from other sources (piped and trucked). Secondary outcomes include perceived water quality, reported uses for rainwater, and observed maintenance outcomes. In addition, we will capture the time and monetary costs of water procurement.

2026-01-15

2027-02-28

Volumes: liters captured via RWH (metered); liters from piped water; liters from trucked water.

Volumes


Liters of rainwater captured via the RWH system (metered): cumulative volume recorded from flow meters installed on the system.
Liters of piped water consumed: aggregated from official water bills or secondary meter readings, standardized to liters for comparability.
Liters of trucked or delivered water consumed: self-reported tanker deliveries (cross-checked with receipts when available) converted to liters.
These outcomes jointly measure changes in rainwater capture and substitution across water sources attributable to the interventions.

Secondary: perceived water quality/reliability, diversity of water sources, maintenance knowledge, and behaviors.
Time and Cost: hours per week devoted to water procurement/handling (collection, treatment, transport, maintenance); monetary expenditures directly related to water (purchases, deliveries, repairs), and observed maintenance outcomes.

The secondary outcome family captures household perceptions, behavioral adjustments, and system maintenance related to rainwater harvesting.
Perceived water quality and reliability: Self-reported perceptions of the clarity, odor, and taste of stored rainwater, as well as the household’s perceived reliability or sufficiency of rainwater availability during the rainy season. These indicators are combined into standardized perception indices for analysis.
Diversity of water sources: Number and share of distinct water sources used by the household (e.g., rainwater, piped water, tanker deliveries, bottled water). This variable reflects the extent of substitution or diversification in household water sourcing.
Maintenance knowledge: Score from survey questions assessing correct understanding of basic maintenance tasks (e.g., cleaning filters, gutters, and first-flush devices, identifying leaks). The measure summarizes technical knowledge about system upkeep.
Maintenance behaviors: Direct observation by surveyors and self-reported actions related to system upkeep (e.g., cleaning frequency, debris removal, and filter or part replacement). These indicators are aggregated into a maintenance-practice index to assess behavioral compliance and engagement with system care.

Time and Costs
This family assesses household resources devoted to water access.
Hours per week devoted to water procurement and handling: total time spent on collecting, treating, transporting, or maintaining water, reported by the main water-responsible household member.
Monetary expenditures directly related to water: self-reported and bill-verified household spending on water purchases, deliveries, and system maintenance or repairs over the relevant reference period.

dor, and taste of stored rainwater, as well as the household’s perceived reliability or sufficiency of rainwater availability during the rainy season. These indicators are combined into standardized perception indices for analysis.
Diversity of water sources: Number and share of distinct water sources used by the household (e.g., rainwater, piped water, tanker deliveries, bottled water). This variable reflects the extent of substitution or diversification in household water sourcing.
Maintenance knowledge: Score from survey questions assessing correct understanding of basic maintenance tasks (e.g., cleaning filters, gutters, and first-flush devices, identifying leaks). The measure summarizes technical knowledge about system upkeep.
Maintenance behaviors: Direct observation by surveyors and self-reported actions related to system upkeep (e.g., cleaning frequency, debris removal, and filter or part replacement). These indicators are aggregated into a maintenance-practice index to assess behavioral compliance and engagement with system care.

Time and Costs


Hours per week devoted to water procurement and handling: total time spent on collecting, treating, transporting, or maintaining water, reported by the main water-responsible household member.
Monetary expenditures directly related to water: self-reported and bill-verified household spending on water purchases, deliveries, and system maintenance or repairs over the relevant reference period.

This randomized controlled trial in Mexico City (N=750) evaluates how engagement and maintenance support, combined with per-liter incentives, affect the use and effectiveness of rainwater harvesting systems. Randomization is stratified by location and baseline RWH storage capacity. The trial has two stages within a single rainy season. In Stage 1 (maintenance support), households are assigned to Control (200), Information + messages (200), Hands-on training (200), or Regular maintenance (150). In Stage 2 (incentives), within each Stage-1 group, 50% of households are additionally assigned to a per-liter cash incentive tied to rainwater captured (counts: Control 100/100; Information + messages 100/100; Hands-on training 100/100; Regular maintenance 75/75 for incentive/no-incentive). Data collection includes baseline, midline, and endline surveys and objective measurement of water capture via flow sensors.

Not available

Randomization will be conducted in an office by a computer.
Randomization is conducted at the household level using stratified random assignment. Households are stratified by location and the size of their rainwater harvesting system's water storage capacity. Phase 1 randomizes participants into four treatment groups (control, information + messages, training, full maintenance). Phase 2 cross-randomizes volume-based financial incentives within all groups.

Household

No

Targeted 750 households

750

Stage 1 (maintenance support), with Stage-2 incentive split shown in parentheses:
Control: 200 HH (100 incentive / 100 no-incentive)
Information + messages: 200 HH (100 / 100)
Hands-on training: 200 HH (100 / 100)
Regular maintenance: 150 HH (75 / 75)
Total: 750 HH (375 incentive / 375 no-incentive)

Power calculations are based on SEDEMA’s study (2021), which measures water volume over the rainy season for a subsample of Rainwater Harvesting Beneficiaries. For baseline, households harvested an average of 10,107.51 liters (SD=6,971) without external maintenance, training, or message reminders. The minimum detectable effect (MDE) is determined by comparing the cost-effectiveness of free maintenance with that of water delivery by tanker trucks. The cost of providing 1 liter via tanker truck is 0.13 Mexican pesos (MXN). One hands-on training visit costs 315MXN and must yield at least 2,423 liters (315/0.13) to be cost-effective: 34% of one SD. After adjusting for take-up, the effect is 2,302 liters, requiring measurement in at least 145 households in this group. The information+messages treatment is expected to be less effective; in particular, we believe that it could obtain 90% of the hands-on training visits effect, that is, 2180.7 liters. A lower amount of liters is not policy relevant if not detected. Adjusting for its 95% take-up rate, the effect is 2,071.67 liters, requiring measurement in at least 179 households. As all these comparisons are done against the control group, the control group must be at least 179 households (per the video + reminders group power calculation). For the full maintenance group, we take a different approach to the power calculation, since we do not expect this treatment to be cost-effective. This treatment is intended to establish an “ideal case scenario” baseline for water quantity. Instead, we assume we can detect 0.5 SD of baseline volume, which is 3,485.5 liters. After adjusting for the take-up rate, this becomes 2,893 liters, requiring 93 households in this group. For the rainwater incentives, this treatment can achieve 0.25 of the SD of baseline volume. Thus, the MDE of this treatment is 1,742.75 liters. We assume a compliance rate of 93.9%, which is the weighted average of the take-up rates for the control group (100%), video+messages (95%), hands-on training (95%), and full maintenance (83.7%). After correcting for the take-up rate, the MDE is 1,637 liters. We require at least 286 households to be assigned to this group.