10.7910/DVN/VIXNNETrejo, GuillermoGuillermoTrejo0000-0003-4286-2808 University of Notre Dame, Department of Political ScienceLey, SandraSandraLey0000-0001-8141-9273CIDE, Political Studies DivisionHarvard Dataverse2019Social Sciencesorganized crimecriminal governancehigh-profile violenceturf warspartisan vertical fragmentationelection cyclesLey, SandraSandraLeyCIDE, Political Studies Division2018-11-012019-06-1826787217424752182567714412302319608056424application/pdftext/tab-separated-valuestext/tab-separated-valuestext/tab-separated-valuestext/tab-separated-valuestext/tab-separated-valuestext/tab-separated-valuesapplication/x-stata-syntaxapplication/pdfapplication/x-stata-syntax1.0CC0 1.0This article explains a surprising wave of lethal attacks by drug cartels against hundreds of local elected officials and party candidates in Mexico, 2007-2012. These attacks are puzzling because criminal organizations prefer the secrecy of bribery over the publicity of political murder. Scholars suggest that war drives armed actors to attack state authorities in search of protection or rents. Using original data of high-profile attacks in Mexico, we show that war need arguments underexplain violence. Focusing on political opportunities, we suggest that cartels use attacks to establish criminal governance regimes and conquer local governments, populations, and territories. We present quantitative and qualitative evidence showing that cartels took advantage of Mexico's political polarization and targeted subnational authorities who were unprotected by their federal partisan rivals. Cartels intensified attacks during subnational election cycles to capture incoming governments and targeted geographically adjacent municipalities to establish controls over large territories. Our findings reveal how cartels take cues from the political environment to develop their own de facto political domains through high-profile violence. These results question the widely shared assumption that organized criminal groups are apolitical actors.This dataverse includes the relevant .do and .dta files for replicating the analyses reported in the manuscript. Please see the ReadMe file for more information on how to use these materials.