New Research Reveals Societal Consequences of Excessive Policing

Locales that generate disproportionate amounts of revenue from low-level fines often have high rates of violent crime

Suzanne McIntosh, CDS affiliated faculty and Clinical Associate Professor of Computer Science, Samuel Smith, Kedar Gangopadhyay, and Simranjyot Singh Gill, of the Courant Institute of Mathematical Sciences, were motivated by the Ferguson, Missouri riots in 2014 to study the influence of city and county revenue generated from fines and forfeitures (primarily from traffic violations) on the incidence of violent crimes.

McIntosh and collaborators used three core datasets: U.S. Census data on tax and fine revenue for over 3,500 reporting jurisdictions (counties and cities) from the Sunlight Foundation; FBI data on violent crimes reported by city; and Stanford Open Policing project data identifying all traffic stops by state annotated with reason(s) the driver was pulled over by law enforcement. They also aggregated and combined U.S. Census 2010 population data and Bureau of Labor Statistics average annual unemployment rates by state for 2013 with their three core datasets. The researchers built on work that examines perceptions of police behavior and a Sunlight Foundation report which found certain states (especially Louisiana, Arkansas, Georgia, Illinois, and Mississippi) generate a disproportionate amount of revenue through fines and forfeitures relative to taxes.

To add insights to the Sunlight Foundation’s report, the researchers ran regressions on their aggregated datasets. First, they tested the common myth (which they note has already been debunked by many academic studies) that higher unemployment rates lead to higher crime rates. Their regression showed no correlation between unemployment and crime rate in 48 states (Hawaii and Wyoming were excluded due to missing data). The second regression, for the same 48 states, found a statistically significant correlation between fines per capita and violent crimes per capita.

Another regression, based on combining statewide stop-and-search data with U.S. census population data, examined how petty crimes are policed for the black population compared to the overall population. This analysis revealed that black drivers in Missouri are 71% more likely to be searched after being stopped compared to the general population, yet the rate at which searches yielded contraband for the black population was equivalent to the rate of the general population. The authors indicate that their results confirm the statistics reported in a travel advisory for Missouri issued by the NAACP, the first of its kind issued for an American state.

Furthermore, their results showed that five of the top eleven states with highest fines per capita also had the highest violent crimes per capita (Missouri, Louisiana, Delaware, Maryland, and Nevada). While the researchers emphasize that “discovery of a statistical relationship between a given factor and violent crime never provides a picture on causation,” they note that for Maryland, while the state has a low unemployment rate and a high percentage of the population has a college education, “compared to the median fines per capita of $1.20 for all cities and counties, Baltimore is 293% higher at $4.73.”

The outcomes of this research demonstrate the need for police agencies and municipalities to reevaluate how they police low-level crimes, and the work shows the efficacy of data analysis as the first step in reforming police practices.

By Paul Oliver