By Valentin Figueroa and Guadalupe Tuñón
According to the US Department of Justice, the rate of imprisonment of African Americans was 938 per 100,000 persons in 2020—five times higher than the rate for whites. The mass incarceration of Black Americans has led activists and scholars to denounce the prison system as “slavery by another name” or “the new Jim Crow”. This growing literature views the carceral state—the police system and courts—as an instrument of racial control. According to this theory, the mass incarceration of African Americans is a result of whites’ efforts to protect their dominant position in the racial hierarchy following increases in African Americans’ political and civil rights over the last century and a half.
In his seminal 1935 book Black Reconstruction in America, W. E. B. Du Bois argued that “the slave went free; stood for a brief moment in the sun; then moved back again toward slavery.” After Reconstruction, Whites enforced a racial hierarchy through a system of legal segregation, the practice of lynching, the perpetuation of negative stereotypes about Black individuals, and a reliance on the police and the courts to disproportionally imprison African Americans.
Research has shown that places in the United States that were more reliant on slavery developed more racist attitudes over time. A more recent strand of work has shown that the expansion of voting rights to African Americans increased incarceration rates and led white Southerners to develop more punitive attitudes regarding the death penalty in the case of murder —as compared to African Americans and whites from the non-South. The mass incarceration of African Americans has been found to decrease the probability of voting in elections.
We make an empirical contribution to this literature through a study of the effect of emancipation on the probability of incarceration. We compare the rate of imprisonment of emancipated and enslaved individuals of African origin in early nineteenth century Buenos Aires. Slavery in urban Buenos Aires was stipendary, meaning that slaveowners forced slaves to sell their labor as artisans and domestic servants. The enslaved could work and spend the night outside their owners’ household, but they were forced to pay their owners a daily stipend, called a jornal. As a result of this particular form of slavery, the enslaved and the emancipated interacted in the same social environment; any differences in the incarceration rates of emancipated and enslaved individuals did not arise mechanically from the fact that the enslaved were segregated in plantations.
Another major advantage of our historical setting is that it offers a unique natural experiment. In 1807, the City of Buenos Aires organized a public lottery to distribute certificates of freedom among slaves who had served as combatants in a prior invasion of the city. This lottery provides random variation in freedman status, which we exploit to identify the effect of emancipation on the probability of incarceration. The lottery took place on the Spanish King’s birthday, November 12th 1807, in the Plaza Mayor, the main public square of Buenos Aires. The names of all the lottery participants were placed in one urn, while a second urn was filled with an equal number of black and white balls (45 white balls represented the number of available certificates of freedom). Names were sequentially drawn from one urn and colored balls from the other. When a name was drawn with a white ball, that slave was freed.
The random assignment of certificates of freedom provides a unique opportunity to experimentally test theories that claim that the expansion of civil rights led to a higher reliance on the carceral state to control newly emancipated persons. Through archival research in Argentina’s national archive, the Archivo General de la Nación, we obtained the full list of lottery winners and the list of enslaved combatants who fought in defense of the city. Figure 1 shows an example of the historical document that recorded the names of the enslaved combatants. To estimate the effect of emancipation on the probability of imprisonment, we manually matched the individuals in our database with police records from 1812-1830 (available from the Archivo General de Policía). To analyze the experiment, we conducted a simple difference-in-means test comparing lottery winners to non-winners.
Figure 2 shows the rate of imprisonment among the individuals who were assigned to emancipation through the lottery (the treatment group) versus those who lost the lottery and remained enslaved (the control group). The intention-to-treat (ITT) results show that emancipation, on average, increased the probability of being imprisoned by 12.1 percentage points. When results are adjusted for some non-compliance with treatment assignment, the complier average causal effect (CACE) is 19 percentage points.
What is the mechanism driving this result? To begin with, it is worth noting that it’s unlikely that emancipated individuals were more likely to run afoul of the law because they spent more time in public spaces. Lottery winners were imprisoned with probability 0.17 and non-winners with probability 0.057, so the ratio of time spent in public spaces across the emancipated and the enslaved would have to be at least 3.12 for this difference to be fully driving our results. This is unlikely because of the “stipendary” nature of slavery in Buenos Aires.
It is also unlikely that our results are an artifact of differential attrition. One could argue that if slaves were more likely to die or migrate out of Buenos Aires, we would artificially observe a higher imprisonment rate for the emancipated for the simple reason that they were more likely to remain in Buenos Aires. To test this claim empirically, we linked the individuals in our experiment with the hand-written records of the 1810 census of the City of Buenos Aires, which we digitized for this project. We find evidence of no differential attrition across experimental groups—treatment status is not significantly associated with the probability of appearing in the census.
What drove the differential incarceration rates across the emancipated and the enslaved? We provide evidence that incarceration against emancipated individuals was driven by arbitrary arrests at the hands of police officers, suggesting incarceration was used as a means of racial control. We disaggregated our results according to the type of charge against incarcerated individuals. We considered three types of charges: (1) robbery, (2) assault, and (3) discretionary charges. The third category encompasses charges that are easy for police officers to trump up, including public drunkenness, profanity, vagrancy, or arrests that fail to specify a charge. We show these results in Figure 3. The effect of emancipation is positive across all three types of charges. However, the magnitude of the effect is biggest when we consider cases of incarceration for discretionary charges (and the estimated CACE is only statistically significant in these cases). We interpret this result as evidence that the higher incarceration of emancipated individuals reflects arbitrary sanctions by police officers.
The main lesson that we draw from our research is that extension of civil rights can be made ineffectual through the systematic incarceration of racial minorities. Emancipated Blacks in colonial Buenos Aires were freed from the private oppression of their masters, but faced a public oppression by the state and its racist criminal justice system. This is one of the central tenets of theories of the carceral state; we bring new quasi-experimental evidence to this burgeoning literature.