top of page

Where's the "systemic racism" in police killings of black civilians?

Is the dice loaded against blacks; or is it the number of throws?

June 14, 2021

A deferential media has reinforced the popular narrative that the killings of Walter Scott, George Floyd, and others are obvious examples of systemic racism in policing. But this is belied by statistical analyses which show that in similar encounters with police, the risk of execution is about the same for blacks and whites. That's why my previous post on policing ignored those notorious cases.

But that can't be! The Washington Post cites statistical proof that cops do prefer to execute black people. For example, 30% of the people killed by police are black, but blacks constitute only 13% of the U.S. population. And on a per-capita basis, blacks are 2.88 times more likely to be killed by police than whites, as shown in column 2 of the table. What more proof do you need??

police killings per capita and per 1000 arrests

The problem is, execution rates based on population (killings per 100K population) are not a valid benchmark for measuring racial bias. That's because they fail to account for the difference in "exposure risk" between black and white civilians. Every uninvited police encounter, from a routine traffic stop to a serious arrest, exposes a civilian to the risk of police violence. The greater the exposure to police intervention, the greater the risk of being killed. Therefore, a more informative measure of the execution rate is killings per 1000 police contacts, or its surrogate, killings per 1000 arrests. (There are no nationwide data for all police contacts, only arrests).

Column 3 shows that blacks are arrested, i.e., exposed to risk, at a rate 2.95 times that of whites. This fact undermines the popular "systemic racism" narrative; it implies that the main reason for the huge racial disparity in death rates shown in column 2 is NOT that a black civilian faces a higher risk of execution in a given police encounter, but that blacks are exposed to the same risk as whites three times more often. (The risk of getting snake eyes on one roll of the dice is 2.8%; on three rolls it's 8.1%).

So, when the execution rate is recalculated as killings per 1000 arrests (column 4), the racial disparity disappears; i.e., the risk of death per police encounter is about the same for black and white civilians.

The purpose of this simplistic statistical exercise is to illustrate the findings of much more sophisticated research. For example, Fryer's study of official police records in Houston and New York found that the risk of execution in a police encounter was about the same for black and whites. Streeter's 2019 study draws the same conclusion.*

Systemic Racism is Evident in Risk Exposure and Proactive Policing.

The findings of Fryer and Streeter debunk just one avenue for systemic racism, but not the other: biased exposure to risk. The excessive targeting of black civilians overexposes them to the risk of police violence, and this results in additional black fatalities. The institutional driver of this racism is proactive policing.

Reactive law enforcement, by contrast, constrains police discretion, which makes exposure to risk less susceptible to racial bias. The prime example of reactive enforcement is arrests for serious crimes such as assault, robbery, rape and burglary. Here the the police are responding to calls from the public - witnesses and victims of crimes. Because of public involvement and the seriousness of the offenses, police have less leeway to inject racial bias in whom they choose to target and arrest.

Those constraints on police discretion are usually missing when enforcement is Proactive. Proactive enforcement is the institutional answer to overcriminalization, especially the criminalization of acts that usually are victimless and private. The mammoth example is possession and sale of illicit drugs, but gambling and prostitution also count.

Those "crimes" do not generate lots of calls and complaints from the public. Since the cops can't rely on the public to generate arrest-leads, they are compelled to be proactive, which means they must drum-up their own opportunities for arrest. Techniques such as undercover stings, "zero tolerance" and Stop & Search generate millions of uninvited encounters which, because of loose police discretion, are disproportionately inflicted on blacks.

For example, a black man's risk of being stopped during NYC's Stop &Frisk program was 4.25 times that of his white counterpart. Blacks are 3.6 times more likely to arrested for pot possession, and 6 times more likely for illegal gambling. On the highways their risk of being pulled over (per 1000 population) is 43% higher than that of whites. The fact is, cities that reduced arrests for minor offenses also saw fewer police shootings.

The point is, even though the risk of execution in a given police encounter is miniscule, millions of excessive exposures are bound to result in superfluous fatalities. The body-count may be small but the headline examples are significant:

  • Most recent (as of this writing) is the case of unarmed Andrew Brown who was shot in the back of the head when he tried to flee a drug bust.

  • The Breonna Taylor case reveals lawlessness in proactive policing. The cops shot her while executing a perjured search warrant based on the uncorroborated suspicion that her apartment was used to stash drugs. No drugs were ever found.

  • And then there was Eric Garner, suspected of selling single cigarettes without a tax stamp. On the theory that such behavior threatened the neighborhood's "quality of life," a band of NYC cops were determined to forcibly arrest him. He died in a choke-hold.

The Garner case raises the question, are the police equally concerned with "quality of life" issues in predominantly white neighborhoods? How about the patriarchal, tight knit neighborhoods of orthodox Jews where child abuse is rarely exposed. Hidden child abuse is certainly more serious than selling single cigarettes, no? Do the cops deploy a special undercover unit to deal with this threat to the neighborhood's "quality of life?"

Proactive enforcement's contribution to the kill rate is not large, but the racial bias is evident. According to MPV data since 2017, only 12% of all homicides stem from traffic stops or drug enforcement. However the racial disparity is exceptionally large. While Blacks represent 32% of all homicide victims, they account for 47% of drug related killings and 40% of traffic stop deaths.

In the above table, 'killings per 1000 arrests' is based only on arrests for serious crimes. When that statistic is recalculated for ALL arrests, which includes the fruits of aggressive proactive policing (e.g., drugs), the effect of racially biased risk exposure emerges. Police killings per 1000 arrests turns out to be 20% higher for blacks.

In sum, the analysis in this post reinforces the theme of my previous post: a focus on headline executions distracts from the fundamental source of systemic racism in law enforcement: the scourge of overcriminalization and its partner, proactive policing. If you want to eliminate superfluous and racist proactive policing, then kill the beast that feeds it - overcriminalization. When the nation saw that prohibition created a crime wave without solving the alcohol problem, it repealed the 18th amendment. Now it's time to repeal the war on drugs.


*The use of Non-lethal violence, on the other hand, is definitely racially biased according to a consensus of research including Fryer's. Why the clear difference between lethal and non-lethal? Fryer finds that racial bias is greatest at the lightest level of physical force and declines as the level of force increases. His explanation is that harsher levels of force attract more scrutiny from higher-ups and the media, which tends to deter bias at at the lethal level.



bottom of page