By Anastasia Wilson and Casey Buchholz
In the wake of the current uprising in support of Black Lives Matter, there has been increasing interest in the use of mainstream empirical methods in economics — like randomized control trials (RCTs) and administrative data evaluation — to address issues of racism and violence in the institution of policing. These interests are well intentioned, but similar to prior debates, we are reminded that “there is reason for concern” about the relevance of these approaches amidst a mass movement calling for deep structural and institutional change. In just two weeks, mass protests have sprung up across the U.S. and the world calling for the defunding, disbanding, and abolition of police as well as the dismantling of white supremacy. This moment has the potential to bring about an institutional and structural shift in our politics, society, and economy. Given this, we will echo many of the concerns shared by economists about the limits of some empirical methods, the biases embedded in administrative data, and the relevancy of these approaches to the current moment calling for immediate change.
Impoverished Economics and the Limits of RCTs
Since the beginning of the Minneapolis uprising, experimental literature on the supposed efficacy of police reform has been gaining attention within the economics discipline. The experimental approach to police reform relies largely on RCTs as well as administrative data evaluation, both methods are applauded for their “objectivity” and in some cases ability to design credible research implementations for causal inference.
The most recent debate over RCTs occurred after the Nobel Prize in Economics was awarded to Esther Duflo along with Abhijit Banerjee and Michael Kremer. As Ingrid Kvangraven explains in her article critiquing the use of RCTs in the field of development economics, the approach assigns specific interventions to a randomly selected group, and then compares how specific outcomes change in the recipient group versus those who did not receive the treatment. For example, with respect to police reform, a popular research design is to randomly select officers to undergo procedural justice training and then compare the likelihood of arrest and use of force in civilian encounters for officers who received training versus those who did not receive training. Similar to how the RCT literature in development economics intends to alleviate poverty through simple interventions, the RCT literature on police reform views the issue of reducing or eliminating police violence simply as a matter of identifying and re-training a “few bad apples” (see, for example, Rozema & Schanzenbach, 2019; and Owens et. al, 2018). While this approach to research “may seem harmless, if not laudable” Kvangraven correctly points out “there are many reasons for concern.”
The RCT approach to police reform that is being promoted is concerned with the activity of individual police officers, not the police as an institution. By design, the RCT approach is rooted in methodological individualism which obscures the roles of institutions and politics, and renders discrimination a matter of individual behaviors and beliefs. In this way, the RCT approach to police reform accepts the institution of policing as a constant and, in doing so, fails to even consider the possibility of institutional transformation as a necessary step towards eliminating police violence. The emphasis on small, technical interventions at the individual-level also loses sight of the ways in which the institution of the police is implicated in broader political and economic dynamics, like anti-black racism, nationalism, militarism, and capitalism. Small interventions may generate seemingly positive results at the micro-level, but do little to challenge the organized forms of power and authority that produce the problems policing simultaneously addresses, creates, and perpetuates as explained too by Sanjay Reddy.
Consider, once again, procedural justice training programs. Procedural justice asks the question “How do law enforcement agencies convince the public that they are benevolent and trustworthy, while making arrests and using physical force?” (see, again, Owens et. al, 2018) The procedural justice model, therefore, focuses on the way in which an officer conducts themselves in the course of taking an action as opposed to the outcome of an action. So, rather than asking why arrests are being made in the first place or why police are given the authority by the state to inflict violence against people who are not allowed to fight back, the RCT approach to police reform naturalizes these conditions and forces us to accept them as immutable.
Methodological Individualism is Blind to Racial Capitalism
The RCT approach to police reform amounts to a perverse exercise in methodological individualism. As discussed above, proponents of the approach argue that the problem of police violence can be solved by correcting cognitive biases and changing the behavioral patterns of individual police officers. However, the RCT approach fails to consider the conditions in which individuals are compelled to make decisions and take action. Individuals don’t exist in a vacuum. Individuals are compelled to act according to the mandate of the institution they serve and, as such, fulfill the purpose of that institution in society. Thus, to paraphrase Karl Marx in the Preface to The Critique of Political Economy, the police officer is a social role that the individual enters into, and, in doing so, dispenses of their individual will. So, then, what is the social role of the police officer? In the United States, policing originated as an institution designed to protect the human property rights of enslavers. Despite the abolition of chattel slavery, the institution of policing has lived on under American capitalism, operating primarily to protect the private property rights of capitalists. It therefore follows that the police officer is required to enforce private property rights in service of the institution of policing.
In fulfilling the social role of the police officer, the personal intentions and attitudes of the individual cease to be relevant. Of course, if called to a retail store to deal with the use of counterfeit money by a customer (as in the case of George Floyd), the officer has the ability to make the decision to not kill the person in question. Sure, the RCT literature shows it is possible to train police officers to kill less or only kill when it is “justified.” That said, the RCT approach fails to consider the fact that enforcing the material insecurity that is inherent to and characteristic of capitalism is still inherently violent. In addition, the RCT approach also fails to consider the question of why police officers have the discretion and power to take someone’s life. Moreover, it is only within the range of violence that spans from depriving someone of the means of survival to depriving someone of life itself that the individual-level decisions and actions of the police officer are able to operate. So, in a material sense, does it really make a difference if the police officer is “respectful” or “courteous” in the course of arresting a shoplifter or drug-user? Empirical methods like RCTs are completely incapable of capturing this range of violence, and, as such, limited to proposing “reformist reforms”: reforms that do not substantially alter the status quo but do legitimize its inherently violent nature.
But, it’s important to take note of the fact that American capitalism is not capitalism in the abstract. American capitalism is racial capitalism. The term, coined by Cedric Robinson in his classic text, Black Marxism, refers to “the development, organization, and expansion of capitalist society” along racial lines (p. 2). Meaning, as Nikhil Pal Singh says, racial capitalism is a specific ordering of capitalism predicated on the “on-going murder” of subjects constructed as Black as well as the “theft of black labor and indigenous lands” (p. 1093). The systematic devaluation of Black life and Black labor is, therefore, essential to the accumulation of capital under racial capitalism. Which is to say that laborers constructed as Black subjects receive less in exchange for equivalent amounts of labor time relative to laborers constructed as non-black subjects in addition to comprising the most precarious sections of the working class by being afforded the least amount of job security and, therefore, by extension, most likely to make up the reserve army of laborers. Taking into consideration the relationship between private property and police violence elaborated on above, it follows that policing is inherently anti-black. Policing under racial capitalism is designed to systematically deny Black subjects the means of survival, and, as such, force Black subjects to participate in a political economic system organized around and animated by their unpaid labor and unwaged existence. The final words of George Floyd and Eric Garner — “I can’t breathe” — articulates the essence of this reality to us in clear terms: racial capitalism is a political economic system organized to deny Black subjects of the ability to live.
In this sense, the RCT approach is, as Jackie Wang would say, completely blind to the economics and “politics of crime data” (p. 247). Crime is not a neutral category. It is a political and economic category rooted in anti-blackness. For instance, in her book, From the War on Poverty to the War on Crime, Elizabeth Hinton details the politics surrounding the construction of the poor Black subject as a criminal, which, once formed, provided the basis for the aggressive expansion of the police state into poor black urban neighborhoods (Hinton, 2016). Thus, returning to Wang, it’s clear that the models used by the RCT approach “are only as good as the datasets they use” and the question of “who collects data and how it is collected” is vitally important [emphasis original] (p. 247). The RCT approach might find benefits from cognitive bias training for police officers working in “high crime” areas, but it is unable to disclose the reasons why an area is so heavily policed and regarded as a “high crime” area in the first place.
Administrative Data is Subject to Power and Bias
What if we think beyond RCTs and towards just administrative data evaluation? As many have pointed out, analysis of administrative data is fraught with issues. First, the data are collected from law enforcement agencies themselves. The data then reflect the power relations of policing and embody such bias. Threads on Twitter pointed out that we need to be careful about what gets recorded as “truth” and research shows specifically that administrative data often masks racial bias in policing. Further, this reliance on so-called objective data collection obfuscates the role, positionality, and gaze of the researcher as well, which too often perpetuate the the multiple racialized and gendered hierarchies of white supremacist capitalism. In fact, these issues of positionality and gaze should be a topic of discussion and questioning within the field of economics. Both of these issues are akin to the feminist critiques raised by Julie Nelson in Feminism, Objectivity, and Economics. Perhaps most importantly though, the timing of this call appears to undervalue and deflect the voices of Black Lives Matter calling for abolition and a complete dismantling and rethinking of institutions that uphold white supremacy, as well as the voices from within economics providing important critiques.
Empirical Methods for Abolition
Our critique is not to say empirical methods are completely useless. Novel approaches to empirical economics can be useful in many contexts, especially when ethically designed or following a big change. Take, for example, Christopher Sullivan and Zachary O’Keefe’s study that analyzes an aberration in NYPD strategy, in which NYPD officers coordinated a work slowdown by limiting foot patrols, criminal summonses, and low-level arrests. The authors find that civilian complaints of major crimes (such as burglary, felony assault, and grand larceny) decreased during and shortly after sharp reductions in proactive policing. The findings of Sullivan and O’Keefe both challenge and disrupt common sense understandings that police deter crime and keep us safe. Thus, empirical methods and data collection can be powerful tools for guiding democratic, transformative, and liberationist struggles so long as they do the work of undermining the forms of organized power and authority the movement is interested in dismantling.
Anastasia Wilson is a PhD candidate in Economics at the University of Massachusetts Amherst. In Fall 2020, she will join the Economics Department at Hobart and William Smith Colleges as an Assistant Professor. She tweets at @anastasiawils.
Casey Buchholz is a PhD student, Department of Economics at UMass-Amherst. He tweets at @caseyrbuchholz.
Photo by Johnny Silvercloud.
One thought on “Abolition Will Not Be Randomized”
[…] Abolition will not be randomized co-written with Casey Buchholz, over on the Developing Economics blog, where we dig into why certain empirical approaches miss the point when it comes to changing the ideology, social relations, and power structures of racial capitalism. “Thus, empirical methods and data collection can be powerful tools for guiding democratic, transformative, and liberationist struggles so long as they do the work of undermining the forms of organized power and authority the movement is interested in dismantling.” […]