Strategic Bureaucratic Opacity: Evidence from Death Investigation Laws and Police Killings

Police accountability is essential to uphold the social contract. Monitoring the monitors is, however, not without difficulty. This paper reveals how police departments exploit specific laws surrounding death investigations to facilitate the underreporting of police killings. Our results show that US counties in which law enforcement can certify the cause of death, including counties which appoint the sheriff as the lead death investigator, display 46% more underreported police killings than their comparable adjacent counties. Drawing on a novel adapted-LATE potential outcomes framework, we demonstrate that underreported police killings are most often reclassified as 'circumstances undetermined' homicides. We also show that law enforcement agencies in counties with permissive death certification laws withhold more homicide reports from the public. The main underreporting results are primarily driven by underreporting of White and Hispanic deaths in our analysis sample, with the effect on Hispanic people particularly pronounced along the US-Mexico border. We do not find that excess underreported killings are associated with more violence directed towards police. We do, however, note a nationwide positive correlation between the permissiveness of gun-laws and underreported police killings. In addition, we find more underreporting in counties which have both high per-capita Google searches for Black Lives Matter and which allow law enforcement to certify the cause of death. Our results do not indicate that other differences in death investigation systems - coroner vs. medical examiner, appointed vs. elected, or physician vs. non-physician - affect the underreporting of police killings.