Leveraging Technology to Prioritize Environmental Enforcement in Bangladesh

How to cite

Mattsson, Martin; Levy, Ro'ee. 2024. "Leveraging Technology to Prioritize Environmental Enforcement in Bangladesh." AEA RCT Registry. AEARCTR-0013521. https://doi.org/10.1257/rct.13521-1.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Martin Mattsson* , Ro'ee Levy

* Primary investigator

Registry ID
AEARCTR-0013521
DOI
https://doi.org/10.1257/rct.13521-1.0
First registered
Last updated
Status
On going
Trial dates
2023-05-01 – 2026-12-31
Countries
Private
JEL codes
D73, H23, K42, Q52, Q53, Q58
Registry record
http://www.socialscienceregistry.org/trials/13521

Abstract

Air pollution is one of the most serious threats to human health in South Asia. Still, environmental regulations are only partially enforced. We test if a dashboard providing bureaucrats with information about the harm of different air pollution sources can direct enforcement toward the most harmful sources. In a nationwide field experiment, we study brick kilns, one of the largest contributing factors to air pollution in Bangladesh. We document the damage of each kiln by coupling satellite data with an air quality model estimating each kiln’s public health effects, create an online tracker summarizing this information, and provide it to the Department of Environment in Bangladesh. We randomize districts into two groups: treated districts, for which the tracker provides information on a random subset of sub-districts, and control districts, for which the tracker does not provide information. We will focus on the following outcomes: enforcement activities, brick kiln owners switching to less polluting technologies, and total health harm. By randomizing both across and within districts, we will test if there are spillovers on areas adjacent to those covered by the tracker.

Intervention

We designed an online dashboard (the “Brick Kiln Tracker”) that allows the Government of Bangladesh to locate all brick kilns and observe the estimated health harm caused by each kiln. We hypothesize that this will allow the government to focus its scarce enforcement resources (funding, bureaucrats' time, as well as political capital) on the most harmful brick kilns. To design the tracker, we use satellite data and machine learning algorithms to identify the type and location of all kilns (Foody et al., 2019). We then estimated the health harm of each kiln, in three stages: First, we estimate the technology-specific emissions of key pollutants from brick kilns following a literature review. Second, we simulate the reaction (with other chemicals), transport, and removal of the pollution in the atmosphere using an air quality model (Thakrar et al., 2022), which takes into account meteorology (e.g. wind and rain), atmospheric chemistry, and pollutant removal. Third, we use maps of population density and epidemiological concentration-response functions (Burnett et al., 2018) to estimate the population exposed and the increased risk of mortality caused by the increase in pollutant concentrations from each brick kiln. Finally, we synthesize this information into a Brick Kiln Tracker displaying the information in a user-friendly and action-relevant way for government bureaucrats.

Experimental design

We randomize what areas of Bangladesh are covered by the Brick Kiln Tracker in a national-scale field experiment covering all of Bangladesh's 170 million population. We randomize which districts will be partially treated by the tracker and which will be "pure control" districts. We then randomize which sub-districts (Upazilas) will be covered within the partially treated districts. This randomization allows us to both measure the direct effect on the treated sub-districts as well as any spillovers due to the reallocation of enforcement resources.

Design details

Randomization method
Randomization done in office by a computer.
Randomization unit
Districts and subdistricts.
Clusters
62 Districts and 322 subdistricts
Observations
For outcomes at the district level: 62 districts For outcomes at the subdistrict level: 322 subdistricts For outcomes at the kiln level measured by satellite data or administrative enforcement data: 7,040 brick kilns For outcomes measured in the kiln survey data: at least 608 brick kilns
Arms
41 Districts (treatment) and 21 Districts (control) 160 subdistricts (treatment) and 162 subdistricts (control)
Primary outcomes
We will focus on the following outcomes: enforcement activities, brick kiln owners switching to less polluting technologies, and total health harm.
Secondary outcomes
See the registered report and pre-analysis plan for details.

Cited by

No citation data available for this registration yet.

Keywords