Causally attributing land-use change from narco-trafficking activities in and around Central America’s protected areas

Keywords: Causal inference, mixed-methods, deforestation, illicit economies
Abstract Type: Paper Abstract

Abstract

The Mesoamerican Biological Corridor is a global biodiversity hotspot, yet forest loss rates there were among the highest in the world over the last two decades. Accelerated deforestation throughout the MBC coincided in space and time with a shift to Central America as the preferred ‘transit zone’ for narco-trafficking, accounting for more than 80% of all U.S.-bound cocaine flows since 2010. Causally attributing that forest loss to direct or indirect impacts of narco-trafficking is difficult due to 1) incomplete, fragments, and/or unreliable data about the location and timing of narco-trafficking activities, and 2) embeddedness and entanglement of illicit with licit economic activities. We have developed a data pedigree system that integrates and makes comparable diverse geospatial data sources, including new media reports, law enforcement data, court records, and remote sensing products. This presentation will describe how this data pedigree is leveraged to utilize the best available data in space and time to causally attribute observed land-use changes to conventional or narco-trafficking activities. Two casual inference strategies, counterfactual land change modeling and quasi-experimental matching, are implemented with the data pedigree and compared on the basis of preliminary estimates of treatment effects (i.e., narco-trafficking presence). This approach has the potential to not only elucidate narco-trafficking’s role in MBC deforestation, but also to advance causal inference methods in land system science more broadly.