Improving worker conditions in Brick kilns: A Randomized Controlled Trial in Bangladesh

Last registered on October 01, 2025

Pre-Trial

Trial Information

General Information

Title
Improving worker conditions in Brick kilns: A Randomized Controlled Trial in Bangladesh
RCT ID
AEARCTR-0016904
Initial registration date
September 29, 2025

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
October 01, 2025, 8:06 AM EDT

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

Last updated
October 01, 2025, 10:04 PM EDT

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

Locations

Region

Primary Investigator

Affiliation
University of Michigan

Other Primary Investigator(s)

PI Affiliation
Stanford University
PI Affiliation
Stanford University
PI Affiliation
UC Berkeley
PI Affiliation
Stanford University
PI Affiliation
Stanford University
PI Affiliation
Stanford University
PI Affiliation
icddr,b

Additional Trial Information

Status
In development
Start date
2025-11-03
End date
2026-07-31
Secondary IDs
Prior work
This trial is based on or builds upon one or more prior RCTs.
Abstract
Bangladesh’s ~7,000 brick kilns contribute substantially to air pollution, responsible for 11% of the country’s particulate matter, 22% of black carbon, and 17% of total annual CO2 emissions . In our earlier work, we developed an intervention, Zigzag 2.0 (ZZK2.0), an intervention offering kiln owners and operators low-cost training and technical support in kiln management and insulation. In a randomized controlled trial with 276 kilns in Khulna Division (2022–2023), 65% adopted the two most important components. Adoption led to 24% less coal use per 100,000 bricks, with 21% reductions in both CO₂ and PM2.5 emissions. At the request of the Ministry of Environment, Forest and Climate Change, icddr,b scaled the intervention across 1000 kilns in 2023-25.

Yet, brick kiln workers represent a particularly vulnerable population (Boyd et al. 2025; Das et al. 2017). Our past study also delivered information to owners on the importance of incetivizing workers to adopt the technical practices, but we found owners did not take up these suggestions. Our recent study also showed about 70% workers reported lack of personal protective equipment (PPE) (Miller 2025), exposing them to risk of burns, head injuries, eye irritation, and smoke inhalation (Shaikh et al. 2012; Sanjel et al. 2016; Das et al.2017). Most lack formal contracts and endure 12–16 hour shifts near 800°C chambers (Maithel et al. 2017), often without shaded rest areas or safe water. Many are seasonal migrants trapped in debt-bondage arrangements, often bringing their entire families, resulting in a high child labor prevalence, estimated in 70% of kilns (Miller 2025). These harsh work conditions cause highly elevated risks of dehydration, heat exhaustion, long-term health complications, and even death (Flouris et al. 2024). At the same time, heat-related illness may lead to productivity losses for owners (Boyd et al. 2025). Other trafficking indicators reported less commonly by workers include wages or benefit witholding (11.8%).

Objectives:
1. Measure uptake of different workplace improvement components and improvements in workplace infrastructure and practices (PPE availability, shaded rest areas, safe drinking water access, breaks during heat periods)
2. Evaluate impact on worker-reported occupational injuries, health outcomes, wage withholding and labor trafficking indicators, and well-being.
3. Assess changes in physiological heat strain indicators (core body temperature, heart rate variability, hydration status).

We will conduct a randomized controlled trial in 200 kilns (120 intervention and 20 control) with brick kilns as the unit of randomization. The study will be implemented across Jashore, Jhenaidah and Satkhira in kilns where icddr,b has established working relationships through previous interventions. Eligible kilns are those that adopted the ZZK2.0 technical intervention in past projects with icddr,b. Intervention kilns will get an integrated training on worker conditions and improvements, as well as a refresher on the ZZK2.0 technical methods. Implementation teams will support kiln owners with planning and implementing the improvements in their kilns through the brick season. Control groups will only be surveyed at baseline and endline.
External Link(s)

Registration Citation

Citation
Ashraf, Sania et al. 2025. "Improving worker conditions in Brick kilns: A Randomized Controlled Trial in Bangladesh." AEA RCT Registry. October 01. https://doi.org/10.1257/rct.16904-1.1
Sponsors & Partners

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

Request Information
Experimental Details

Interventions

Intervention(s)
Owners of kilns in the intervention arm will receive an integrated technical and work conditions training package (ZZK2.0+) which includes a refresher training on the proven Zigzag 2.0 kiln efficiency improvements combined with a worker condition improvement (WCI) package designed to enhance occupational safety, improve payment practices, and improve living conditions for kiln workers. The WCI package, finalized during our pilot, is structured around five components: (1) provision of personal protective equipment (PPE) and occupational safety measures, (2) timely wage payment monitoring, (3) improvements to workplace and living conditions, (4) heat mitigation strategies, and (5) child labor awareness and prevention.
Intervention Start Date
2025-12-21
Intervention End Date
2026-02-22

Primary Outcomes

Primary Outcomes (end points)
1. Personal Protective Equipment (PPE) Index
2. Workplace conditions
3. Timeliness of wages
4. Labor trafficking indicators
5.Heat stress
Primary Outcomes (explanation)
1. Personal Protective Equipment (PPE): We use principal components analysis (PCA) to calculate a continuous PPE index for each worker, using a set of 9 indicators taking a value of 1 if a given worker reports access to a given PPE item (cooling towels, gloves, helmets, head coverings, boots, goggles, masks, first aid support, oral rehydration salt (ORS)), and zero otherwise. Our study is powered at 80% to observe a minimum decline of 0.29 in a PPE index computed using previous season data (standardized effect size of 0.27), with a relatively high ICC of 0.24.
2. Workplace conditions: We use principal components analysis (PCA) to calculate a continuous index reflecting the quality of working conditions each worker reports. Components include sanitation conditions, access to rest areas, cooling sheds, and amenities in living quarters. Our study is powered at 80% to observe a minimum decline of 0.44 in a working conditions index computed using previous season data (standardized effect size of 0.28), given a high leval of intracluster correlation (ICC = 0.39).
3. Timeliness of wages: For each kiln, timeliness of wages is measured using share of workers reporting on-time payment during the preceding 3 months of work (or since the beginning of the season, if shorter than 3 months). Although available data for the timeliness of wages is limited, we estimate a minimum detectable effect size of 7.15 percentage points in the probability of timely pay (a standardized effect size of .22 relative to baseline), given a relatively high ICC of 0.32.
4. Labor trafficking indicators: Conditions contributing to labor trafficking are measured using a standarized set of indicators developed by the United States Department of State Trafficking in Persons Office. The standard tool consists of 39 total indicators (4 extremely strong indicators, 14 strong indicators, and 21 medium indicators) for conditions related to labor trafficking determination. Our survey focuses on a subset of 17 of these indicators deemed culturally appropriate and relevant to the intervention. Our primary outcome is then computed as a continuous measure of the total number of indicators experienced (weighting medium indicators at 0.66). Previous surveys collecting similar data show that our study is powered to detect declines of 0.342 trafficking indicators (an 18% decline; standardized effect size of 0.26) with 80% power at a 5% confidence level.
5. Heat stress: We measure heat stress following standard clinical practice using maximum heart rate, skin temperature, and urine specific gravity. Our study is powered at 80% to observe a minimum decline of 5.43BPM in maximum heart rate (standardized effect size of 0.34), 0.49° in skin temperature (standardized effect size of 0.301), and 0.0022 in USG (standardized effect size of 0.337), assuming a relatively high intracluster correlation of .3 (estimated using data on outcomes related to work conditions measured in a preceding study of the same kilns).

Secondary Outcomes

Secondary Outcomes (end points)
Worker reported health outcomes
Secondary Outcomes (explanation)
Respiratory outcomes: Respiratory health will be assessed through worker-reported symptoms (cough, shortness of breath, chest tightness, wheezing) using structured recall periods of 2 days, 1 week, and 2 weeks prior to survey administration. Symptoms will be scored using frequency categories (none, mild, moderate, severe) and functional impact on work activities. Given the seasonal nature of brick kiln operations and potential recall bias with longer time periods, the 2-day and 1-week windows will serve as primary respiratory endpoints. Based on occupational health studies in similar dusty work environments, we estimate our study has 80% power to detect a 15-20% reduction in moderate-to-severe respiratory symptom reporting (standardized effect size of approximately 0.25), assuming an intracluster correlation similar to our other worker-reported outcomes (ICC ≈ 0.30).

Injuries: Workplace injuries will be measured using worker recall of incidents occurring in the past 2 weeks, categorized by type (cuts, burns, falls, crushing injuries), body part affected, and severity (required stopping work, sought medical care, lost work days). The 2-week recall period balances accuracy of reporting with sufficient observation time to capture injury events. Our sample size provides 80% power to detect a reduction of approximately 8-12 percentage points in injury incidence (standardized effect size of 0.22-0.28), based on baseline injury rates observed in comparable brick manufacturing settings and assuming moderate intracluster correlation (ICC ≈ 0.25).

Experimental Design

Experimental Design
We will use our database of past enrolled kilns to create a list who have adopted ZZK2.0 in the Jashore, Jhenaidah and Shatkhira districts of Khulna Division, Bangladesh (n = 200). Field teams will approach these kilns, secure their consent to participate in the trial, and conduct a baseline survey. From the enrolled kilns, we will stratify the kilns by phase of prior project participation and randomize kilns into one of 2 groups: intervention (n=120) and control (n=80).
Experimental Design Details
Not available
Randomization Method
Stratified randomization done in an office on a computer.
Randomization Unit
Unit of randomization is a brick kiln, all workers within a kiln are treated, as the training is delivered to the kiln owners and managers.
Was the treatment clustered?
Yes

Experiment Characteristics

Sample size: planned number of clusters
200 brick kilns
Sample size: planned number of observations
1000 kiln workers
Sample size (or number of clusters) by treatment arms
120 brick kilns in the treatment arm, 80 brick kilns in the control arm.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
Our study is powered at 80% to observe a minimum decline of 0.29 in a PPE index computed using previous season data (standardized effect size of 0.27), with a relatively high ICC of 0.24. Our study is powered at 80% to observe a minimum decline of 0.44 in a working conditions index computed using previous season data (standardized effect size of 0.28), given a high leval of intracluster correlation (ICC = 0.39). Although available data for the timeliness of wages is limited, we estimate a minimum detectable effect size of 7.15 percentage points in the probability of timely pay (a standardized effect size of .22 relative to baseline), given a relatively high ICC of 0.32. Our study is powered to detect declines of 0.342 labor trafficking indicators (an 18% decline; standardized effect size of 0.26) with 80% power at a 5% confidence level. Our study is powered at 80% to observe a minimum decline of 5.43BPM in maximum heart rate (standardized effect size of 0.34), 0.49° in skin temperature (standardized effect size of 0.301), and 0.0022 in USG (standardized effect size of 0.337), assuming a relatively high intracluster correlation of .3 (estimated using data on outcomes related to work conditions measured in a preceding study of the same kilns).
IRB

Institutional Review Boards (IRBs)

IRB Name
IRB Approval Date
IRB Approval Number