Rohingya Refugee Camps and Forest Loss in Cox’s Bazar, Bangladesh: An Inquiry Using Remote Sensing and Econometric Approaches

Nfamara K. Dampha, Colette Salemi, and Stephen Polasky

World Bank Policy Research Working Papers, No. 9948


More than 740,000 Rohingya refugees from Myanmar fled to Bangladesh in late 2017 and were accommodated in large refugee camps in Cox’s Bazar district. This paper evaluates whether the Rohingya refugee camps in Bangladesh have caused changes in forest cover in the surrounding areas and investigates the specific causes of environmental impacts. The authors distinguish between: (1) direct impact of either clearing land for the expansion of the camps or harvesting of forest products by refugees for shelter and cooking fuel; and (2) indirect impact of the camps on local economic activity, infrastructure and services, which in turn might attract Bangladeshi people to settle in the vicinity of the camps.

For the econometric analysis, the authors randomly selected 10,000 locations across Cox’s Bazar, and compared locations within five kilometers of the camps (the ‘treatment’ group) with locations 10 to 15 kilometers from the camps. The authors control for a variety of fixed effects such as soil fertility, whether the locations are protected environmental areas, steepness of the terrain, and seasonal variations in vegetation cover.

The analysis draws on several data sources, including: (1) satellite imagery from the US Geological Survey, to classify locations according to land use and to assess levels of vegetation cover; (2) Protected Planet’s database of protected areas to classify locations as protected or unprotected areas; (3) annual population data from Worldpop; (4) nighttime lights data produced by the National Oceanic and Atmospheric Administration, which are used as a proxy for economic activity; and (5) Google Maps for the locations of main roads. Descriptive data suggest that forest area has declined since 2010 in the study region, and that forest cover loss has accelerated since 2017 compared to earlier years.

Main findings from the empirical analysis:

  • There has been a decline in forest cover and an expansion of settlement areas within 5 kilometers of the refugee camps. Following the opening of the camps, locations within 5 kilometers of camp boundaries were 7.6 percentage points more likely to become deforested and 13 percentage points more likely to be classified as a settlement, compared to areas further away.
  • Two thirds of forest losses occurred between 1 to 5 kilometers of the camps, driven by settlement expansion, and forest losses are concentrated outside protected areas. Within 1 kilometer of the camp, forest cover has declined—most likely due to the activities of camp residents—but settlement areas have not expanded in these locations. However, between 1 to 5 kilometers of camp boundaries, forest cover has declined, and settlement areas have expanded, suggesting that forest losses were driven by the expansion of settlement areas, most likely due to the in-migration of Bangladeshi citizens. Excluding locations within one kilometer of a camp boundary, locations between one to five kilometers of the camp boundaries were 4.5 percentage points more likely to become deforested and 11 percentage points more likely to be classified as a settlement, compared to areas further away.

The authors argue that the causes of deforestation following the opening of the camps go beyond the direct impact of refugees. The camp and its residents appear to have contributed directly to forest losses within one kilometer of the camp. However, within one to five kilometers of the camp, where forests are being cleared for settlements, it is likely that Bangladeshi citizens moved towards the camps in search of economic opportunities and/or improved services due to the presence of the camps.