Can demining programs decrease conflict and deliver economic and social benefits—even in active conflict zones? And are the benefits of landmine clearance larger or smaller in places with more or less conflict?
In a recent report, AidData researchers tackle these important questions for the country with one of the world’s worst landmine burdens: Afghanistan. They have finished the third phase of a geospatial impact evaluation of the Mine Action Programme of Afghanistan (MAPA), a national effort to remove land mines. The report was commissioned by the global development organization Itad, as part of its contract to provide monitoring and evaluation services to the UK’s Foreign, Commonwealth and Development Office's Global Mine Action Programme 2. The third phase of the project sought to understand the role of conflict in mine action and the role of mine action in subsequent conflict; a second goal was to further investigate how landmine clearance affected aid distribution, economic growth, land use, and trust in government.
The research team found that landmine removal had broad positive impacts on outcomes ranging from economic growth and trust in government, to conflict levels and land use. They also uncovered that the baseline level of conflict in an area that was later cleared had interesting differential effects on certain outcomes.
“The landmine clearance strategies employed in Afghanistan from 2008 onwards led to increases in nighttime light levels (NTL) visible in satellite imagery. NTL are a good proxy for economic growth, meaning landmine clearance very tangibly resulted in improved economic livelihoods. What’s even more interesting is that the results were stronger for areas that had experienced more conflict at the study’s baseline,” said Rachel Sayers, a Research Scientist at AidData and a co-author of the evaluation. “We also found that landmine clearance is associated with increased aid at the district level, potentially by making areas safer to operate and site projects in. Districts with more clearance activity were more likely to receive foreign aid a year later, and the effect was even stronger two years later. Furthermore, this effect isn’t being driven by outlier aid projects; it remains even if we remove large projects to focus on small ones,” said Sayers.
“The takeaway is clear,” said Dr. Ariel BenYishay, AidData’s Chief Economist and Director of Research and Evaluation and a co-author of the evaluation. “Landmine clearance in Afghanistan had broad positive impacts, on everything from economic growth and productive land use to citizen trust in government. But, depending on the results that program planners want future mine clearance efforts to have, they should keep an eye toward baseline levels of conflict in areas.”
Far from being a twentieth-century relic, landmines maim or kill more people in Afghanistan than anywhere else in the world. MAPA has successfully cleared more than 18 million landmines from Afghanistan since its founding by the UN Mine Action Service (UNMAS) in 1989. Yet until AidData was commissioned to perform an impact evaluation of MAPA, little was known about the program’s long-term effects on economic growth, land use, and conflict patterns.
AidData’s research into MAPA’s impacts is timely, coming in the wake of US withdrawal and subsequent Taliban takeover nearly two years ago. Originally, the research team planned to be on the ground in Afghanistan as part of the project’s third phase, in order to collect ground-truth data to train machine learning algorithms to generate additional outcome measures. Unfortunately, this did not come to fruition, due to the US’s withdrawal from Afghanistan and the subsequent Taliban takeover.
The data the research team used for this series of evaluations ended in 2020, so they have no data on landmine clearance after the Taliban returned to power in August 2021. Even so, the research reveals the complex interaction between conflict and mine action, as well as the potential for demining to improve economic development in ongoing conflict zones.
“Although our results don’t speak directly to mine action after the Taliban takeover of Afghanistan, the trends we identify provide insight into how the impacts of mine action might change as conflict dynamics in the country continue to evolve,” said Madeleine Walker, a Junior Data Analyst at AidData and a co-author of the evaluation. “Another key lesson learned from this project is how well geospatial impact evaluations (GIEs) like this one work to evaluate a large program in an active conflict zone. GIEs are an incredibly valuable tool to deliver insights to program managers, even when on-the-ground access is impossible,” said Walker.
Despite the challenges, the research team was able to arrive at robust conclusions drawing from a wide array of data—including administrative data on the locations and timing of landmine clearances and aid distributions; remotely sensed data, including daytime satellite imagery and nighttime lights (NTL) of cleared and comparison areas; survey data on trust in government; and event data on the locations and timing of conflicts.
“The impact evaluation required merging administrative, remotely sensed, survey, and event data through a variety of statistical programs,” said Walker. “I worked on calculating how many conflict events were present in each polygon of land area, which alone took more than a day to run,” explained Walker. The team also employed cutting-edge machine learning techniques to help them analyze the data. Dr. Kunwar Singh, a Geospatial Scientist at AidData and co-author of the evaluation, trained algorithms to process satellite imagery of contaminated and cleared areas to determine the land uses of each area, including built-up, farmland, and grassland.
The culmination of this process was an analysis that painted two clear images of mine action in Afghanistan. First, mine action and conflict were inextricably linked. The government authorities directing mine clearance in Afghanistan appear to have selected sites for clearance based, in part, on pre-existing conflict levels, so clearance tended to happen after conflict had decreased or stabilized at lower levels. After controlling for pre-trends in conflict, however—in order to isolate the direct impacts of mine action—it was evident that mine action from 2008 onwards reduced subsequent conflict in and of itself.
Second, mine action from 2008 onwards had an overwhelmingly positive impact on a wide range of indicators of economic development across a wide range of geographic locations. These impacts were seen in both electrified and unelectrified areas, and in areas with both high and low baseline conflict.
“If one thing is clear from our research, it’s that, while policymakers must monitor conflict in order to target clearance activities most effectively, mine action can have strong positive impacts even in active conflict zones,” says Sayers.
