Policing and race: The role of electoral accountability

Giovanni Facchini, Brian Knight, Cecilia Testa 07 July 2020



On 25 May 2020, George Floyd, an African American man accused of using counterfeit currency, was killed during an arrest in Minneapolis, Minnesota. On 12 June, another black man, Rayshard Brooks, was found asleep in his car in a restaurant driveway in Atlanta, Georgia. A police officer shot and killed Brooks as he was trying to escape. These and other incidents of police brutality against black Americans have led to widespread protests, with huge crowds demanding changes to policing practices and greater public accountability for police forces even in the midst of a global pandemic.

The relationship between race and policing has been extensively studied in the academic literature. Traditionally, attention has focused on the racial composition of police forces, which have historically been dominated by white officers. Extant research has examined whether diversifying police forces leads to a shift in arrest patterns according to race (Donohue and Levitt 2001, McCrary 2007, Bulman 2019). A related literature has also studied whether the differential treatment of black Americans by police officers represents statistical or preference-based discrimination (Knowles et al. 2001, Antonovics and Knight 2009, Anwar and Fang 2006).

While progress has been made in understanding the effect of the racial composition of both the police force and its leadership on the treatment of black citizens, much less is known about the role of elections and political accountability, a relationship recently emphasised by former US President Barack Obama. This link is important for at least three reasons. First, many chief law enforcement officers (CLEOs) in the US – essentially all sheriffs in the South and some municipal police chiefs – are directly elected rather than appointed. Second, CLEOs are in charge of policing practices with immediate effects on the treatment of minorities and are in a position to affect departmental culture through their leadership. Third, black Americans’ voting rights have a troubled history: first granted after the Civil War, then restricted during the Jim Crow era in the South, and finally re-established during the 1960s. Even today, significant restrictions on voting result from, for example, voter ID laws and the disenfranchisement of felons, both of which are believed to disproportionately impact black voters.

In a recent paper (CEPR DP 14946), we study the role played by CLEO accountability to black voters by exploiting the dramatic change in voting rights brought about by the 1965 Voting Right Act (VRA), which compelled the covered Southern jurisdictions to remove restrictions on the franchise and seek federal authorisation for any subsequent changes to their voting laws. As a result, black voter registration and voter turnout soared, as documented by Cascio and Washington (2014) and Wright (2013). The existence of segregated neighbourhoods implies that police treatment of blacks was an important policy area in which white politicians could take steps to improve the lives of black voters without necessarily generating a backlash from the white electorate.

To examine these issues, we use county level data on the patterns of arrests by race from the FBI Uniform Crime Reports spanning the period 1960-1980. If coverage has an effect, areas with larger shares of African Americans should experience a decline in their arrest rates compared to those of whites. However, given that the racial composition of arrests in those areas might have changed for reasons other than federal intervention, the comparison of arrest patterns by race before and after the VRA is combined with the addition of a control group, including jurisdictions of the former Confederacy that were not covered by the VRA. Thus, our analysis compares: 1) arrest patterns for black versus white suspects, 2) before and after the VRA, 3) in covered versus non-covered areas, and 4) in counties with large black populations, which are more impacted by the VRA, versus counties with lower numbers of African Americans.

Figure 1 Arrest rates by race and coverage status

Our baseline results are illustrated in Figure 1, where we relate the long-run difference in arrest rates (pre- versus post-VRA) to the share of black Americans in 1960. Using bin scatter diagrams, we separately consider covered and non-covered counties, and black and white arrests. As shown in the left panel, black arrest rates increase more rapidly in areas with larger shares of African Americans in both covered and non-covered counties (see e.g. Forman 2017). But, importantly, the growth is less pronounced in covered areas, consistent with our hypothesis that enfranchisement leads to better treatment of black residents by police, and especially so in counties with larger shares of African Americans. White arrest rates, by contrast, appear to be independent of the black population in both covered and non-covered areas (see right panel). In additional analysis using detailed information on the nature of the alleged crimes, we find that our results are driven by less serious (misdemeanour) offenses – over which police have more discretion – rather than more serious felonies.

We then turn to explore potential mechanisms. We start by further investigating the electoral channel, analysing whether our findings differ depending on the rules by which CLEOs are selected. We document that our baseline results are driven by arrests carried out by sheriffs (who are always elected) rather than by police chiefs (who are typically appointed). Furthermore, exploiting variation in local statutes and charters prescribing the appointment rules for police chiefs, we find that the aggregate pattern is driven by counties where police chiefs are exclusively appointed. By contrast, in the presence of elected police chiefs, arrests of black suspects follow patterns similar to those uncovered for sheriffs. All of these results provide prima facie evidence that the VRA led to changes in police behaviour by increasing the accountability of CLEOs to black voters.

We also uncover evidence indicating that such improvements are driven by a change in the practices developed by CLEOs, rather than the race of the officers or other elected officials. In particular, we document the near complete absence of black CLEOs even twenty years after the passage of the VRA. This allows us to rule out the idea of an identity politics channel for sheriffs or police chiefs. As for other elected officials, we find that covered counties with larger pre-existing shares of African Americans do elect more black county commissioners, but only in the presence of single member districts. Importantly, these areas do not experience an improvement in the treatment of their black populations, which is instead recorded in covered counties that do not elect more black commissioners. Hence, the race of other local office holders does not appear to play an important role either. Lastly, investigating the racial composition of the police force, we find that while the South experienced an overall increase in the number of black police officers, there is no difference between covered and non-covered counties.

We then turn to four alternative explanations for our findings. First, in the debate surrounding the recent killings of George Floyd and Rayshard Brooks, much attention has been devoted to the role played by police unions, which have often been accused of pursuing aggressive tactics to protect their members, curbing the CLEO’s ability to discipline their workforce in case of misconduct. Second, the observed improvement in the racial pattern of arrests could be driven by improvements in schooling, labour market conditions, and more generally socio-demographic characteristics of African Americans. Third, crime patterns might respond to changes in policing patterns following the VRA. Fourth, our results could be driven by elevated arrests of black protestors during the pre-VRA era. We address each of these in turn and show that our results, especially those involving differences in arrest patterns between elected and appointed officials, are driven by none of these alternative explanations.

Taken together, our analysis indicates that the enfranchisement of minority groups can lead to improved treatment by police, but only when CLEOs are elected. While historical in nature, our findings have significant policy implications, especially given the ongoing debates over race, policing, and voting. On the one hand, although black Americans continue to be disproportionately targeted by law enforcement, our results indicate that the election of CLEOs affects accountability and improves police treatment of minorities. On the other hand, since the franchise matters, if recently enacted changes in the costs of voting – such as voter ID laws – have disproportionate effects on black voters, this might worsen their treatment by police.


Antonovics, K and B G Knight (2009), “A new look at racial profiling: Evidence from the Boston police department”, The Review of Economics and Statistics 91: 163–177.

Anwar, S and H Fang (2006), “An alternative test of racial prejudice in motor vehicle searches: Theory and evidence”, American Economic Review 96: 127–151.

Bulman, G (2019), “Law enforcement leaders and the racial composition of arrests”, Economic Inquiry, 57: 1842–1858.

Cascio, E U and E Washington (2014), “Valuing the vote: The redistribution of voting rights and state funds following the voting rights act of 1965”, The Quarterly Journal of Economics 129: 379-433.

Donohue, J J and S Levitt (2001), “The impact of race on policing and arrests”, Journal of Law and Economics 44: 367–394.

Forman, J Jr. (2017), Locking up your own: Crime and punishment in black America, New York: Farrar, Straus and Giroux.

Knowles, J, N Persico and P Todd (2001), “Racial bias in motor vehicle searches: Theory and evidence”, Journal of Political Economy 109: 203–229.

McCrary, J (2007), “The effect of court-ordered hiring quotas on the composition and quality of police”, American Economic Review 97: 318–353.

Obama, B (2020), “How to Make this Moment the Turning Point for Real Change”, Medium, 1 June.

Wright, G (2013), Sharing the Prize, Cambridge, MA: Harvard University Press.



Topics:  Politics and economics Poverty and income inequality

Tags:  Racism, policing, US, elections, American South

Professor of Economics, University of Nottingham and University of Milan and CEPR Research Fellow.

Professor of Economics, Brown University; and Research Associate, NBER

Professor, School of Economics, University of Nottingham


CEPR Policy Research