Abstract
This study presents the first ever comparative regional portrait of racial inequality and incarceration across the Americas, using census data from Brazil, Cuba, Puerto Rico, and the United States mainland. While racism is known to pervade criminal justice across the US mainland, Latin American prisons remain understudied, with the entire region often construed as racially harmonious and uniformly “mixed” rather than racially plural or stratified. Our findings reveal consistent underrepresentation of white individuals and overrepresentation of Black individuals in all countries. Mixed-race individuals in Brazil, Cuba, and Puerto Rico experience higher incarceration rates than whites but lower rates than Blacks. These findings challenge the conception that the US mainland is unique in its historically entrenched profile of structural racism, while highlighting varying degrees of racial inequality internationally. Whereas Cuba and the US mainland display relatively higher levels of racial inequality in imprisonment, Puerto Rico and Brazil display relatively lower levels.
Introduction
With 1.5 million people behind bars—twice as many as twenty years ago—Latin America has superseded the United States as the region with the fastest growing prison population worldwide (World Prison Brief, 2021). Unlike the United States, however, the racialization of Latin America's prisons remains unclear. With the notable exception of Brazil (Departamento Penitenciário Nacional, 2016), Latin American countries tend not to monitor the ethnoracial makeup of their prison populations, hindering the development of regionally nuanced racial critiques of incarceration.
Using internationally standardized census data, we provide the first ever comparative portrait of racial inequality across prisons in the Americas, focusing on Brazil, Cuba, Puerto Rico, and the United States mainland. While such comparative knowledge is critical to advancing racial justice and countering incarceration, comparing the racial dynamics of imprisonment across the Americas poses significant methodological challenges. It is not simply that “race” is conceived in incredibly diverse ways across national contexts, with racial categories and their cultural meanings varying widely between nation states (or, in the Puerto Rican-US context, varying considerably within the same country, Duany, 2005). Even more challengingly, decades of race research in Latin America points to the complicating factor of Mestizaje (racial mixture). This is an old discursive tradition encompassing at once an ideology, a governing strategy, and a regime of representation that has worked over the course of two centuries to render most Latin Americans mixed, few categorically Black, white, or Indigenous (Wade, 2005).
Though race theorists have vigorously de-constructed Mestizaje (Golash-Boza and Bonilla-Silva, 2013), with studies of “silencing” and “pigmentocracies” arguing that it conceals the fact that those with darker complexions occupy the most vulnerable rungs of society (Rodríguez-Silva, 2012; Telles, 2014a, 2014b), when it comes to understanding racism across Latin American criminal justice systems, carceral researchers are confronted by precisely this problem of invisibility (LeBrón, 2019: 13; Weegels, 2020: 312). Explicit racial analyses of incarceration in Latin America are scarce. Instead, Latin American prison studies has devoted the lion's share of its attention to analyzing how national prison booms map onto neoliberal policy projects and reinforce class hierarchies (Azevedo and Cifali, 2016; Grajales and Hernández, 2016; Sozzo, 2016) and to exploring the informality of Latin American prison governance (Antillano, 2017; Darke, 2017; Dias and Salla, 2017).
This raises a question, a contention, and a missed opportunity. In that order. First, the question: Do Latin American prison regimes harbor racial inequalities, and if so, are these racial inequalities comparable in any way to those of the United States mainland? Does it make sense to speak of “Latin American” or even an “international” racial profile of imprisonment, that some scholars are characterizing as a “transnational logic of Black disposability” (Alves, 2016)? Or are Latin American nations so different from each other, and/or so different from the United States mainland that there is no singular or coherent racial logic to imprisonment in the Americas? If the latter is true, what new avenues of research are needed to understand locally distinct profiles of racial inequality as they differentially manifest across the Americas?
Two strongly counterposing reflexes—which is where the contention comes in—generally greet such questions. A common reaction espoused by Latin American politicians (Castro, 1991; Muñoz Marín, 1925), Puerto Rican high school textbooks (Godreau et al., 2008), and (intriguingly) French sociologists Pierre Bourdieu and Loïc Wacquant (1999) is that structural racism is a problem unique to the United States, one that does not extend to Latin America. These sociologists have even gone as far as denouncing the whole project of documenting racism in Latin America as an “imperialist” endeavor that imposes the United States’ rigid system of Black-white racial segregation onto more racially integrated nations and achieves little more than “symbolic violence” (Bourdieu and Wacquant, 1999: 41–46). The notion that the US mainland is unique in its historically entrenched profile of structural racism is strongly challenged by recent evidence that racial disparities may be greater in some Latin American countries than on the US mainland. The seven-year gap in life expectancy between Afrodescendent and non-Afrodescendent men (64.9 vs. 71.9) in Cali, Colombia, for example, is notably higher than the equivalent 5.8-year gap in the United States (71.8 vs. 77.6, see Urrea-Giraldo, 2012). The 55–60% wage gap between Black and white Brazilian men is higher than the 40–45% wage gap between Black and white American men (Telles, 2014a, 2014b), while in Mexico, a greater proportion of indigenous people live in poverty relative to the general population (70.3% vs 38.7%), compared to the equivalent US figures (25.4% of indigenous people vs. 12.3%, see CONEVAL, 2016).
A counterposing (and unanimous) reaction from Race and Ethnicity Studies, African American Studies, and scholars of the African diaspora is that yes, racism is endemic in Latin America, but it does not assume the same formation regionally. But here is where things become far less unanimous. Studies of racial inequality across Latin American nations have begun to compare ethnoracial disparities internationally across socioeconomic indicators such as employment, home ownership, health, and wealth (Ayala-McCormick, 2021; Caraballo-Cueto and Godreau, 2021; CEPAL, NU, 2020; Solís, 2017; Telles, 2014a, 2014b). These studies undeniably paint a much more complicated and heterogeneous racial portrait than US audiences are accustomed to. A key finding from this literature is that there are not simply different kinds of racial inequality in Latin American countries (e.g., among Indigenous and Mestizo populations verses Black and white populations) but also that degrees of racial inequality vary widely too. In Brazil, for example, the median Black woman (in terms of her income) makes just 51% as much as the median white man. On the US mainland, the equivalent figure is 58%, while in Puerto Rico, the median Black woman makes 96% as much as the median white man (Ayala-McCormick, 2021). This emerging portrait of racial inequality is providing information that is desperately needed for advancing racial justice. Yet no study to date has compared incarceration and racial inequality across the Americas.
Despite claims to race's “invisibility”—and here is where the missed opportunity comes—a “race” or “ethnicity” question of one variety or another can now be found in the national censuses of every nearly every Latin American country (the only exception being the Dominican Republic, Loveman, 2014). While census administrators will sometimes omit prisons entirely or neglect to gather information about whether individuals who complete the census are incarcerated, thereby burying the incarcerated population within the general count, a number of countries that extend the census to prisons also record information about the “living arrangement” of census participants. As we detail below, this means that it is possible to use national census data pertaining to “living arrangements” to approximate the incarcerated population and, having refined this approximation, it is possible to measure ethnoracial inequalities among incarcerated populations for the different countries and/or territories and to begin to construct a preliminary comparative portrait of racial inequality across the Americas.
The following analysis summarizes the diverse profiles of racial inequalities that manifest among incarcerated populations in four sites with large Afrodescendent populations: Brazil, Cuba, Puerto Rico, and the United States mainland. But before we elaborate our choice of sites and describe these patterns of racial inequality, let us say a bit more about what is currently known about race and incarceration in Latin America and explain why that understanding is so limited.
Counting race in the Americas
The United States (the mainland, not the colonies) stands out as the only place in the Americas where race has stayed at the forefront of public debate about incarceration since the abolition of slavery. State practices of ethnoracial counting were critical to settler colonialism and racial slavery during the seventeenth and eighteenth centuries (Levitan, 2011), two forms of “racialized accumulation by dispossession” (Wang, 2018: 99) upon which the nation was founded in 1776. When slavery was replaced by Jim Crow laws and practices that criminalized African Americans, emerging evidence for relatively higher incarceration rate for Black populations soon gave rise to a “statistical discourse” (Muhammad, 2021: 4) about Black crime in the political imaginary. In its nineteenth and early twentieth century iterations, this statistical discourse attributed African Americans’ elevated incarceration to an inherent Black criminality and inferiority. Over the course of the twentieth century, statistical knowledge about racial difference was put to new progressive purposes in the context of Black social movements fighting for civil rights (Prewitt, 2012). Racial statistics previously geared towards perpetuating myths of white superiority were redeployed to expose and critique the racism of US society. Gradually, ethnoracial counting became a progressive governing strategy for monitoring adherence to civil rights and equality legislation as well as one of the most important prisms for understanding racism. On the US mainland then, ethnoracial counting has played a critical role both in enforcing racial subjugation and in advancing anti-racist efforts, which is why, today, not just shelves but libraries worth of volumes are devoted to counting, modelling, and theorizing the disproportionate incarceration of Black Americans (Hinton and Cook, 2021).
In Latin America, post-independence, there was a swift rejection of the racial caste systems enforced under European colonialism (Andrews, 2016: 85–116). Embracing ideologies of Mestizaje, many nations suspended or drastically reduced the practice of gathering racial information, considering it emblematic of colonial oppression. While countries had varying histories of racial counting (Loveman, 2014: 208), it remained a marginal practice confined to census offices and was not widely adopted across governmental divisions. Before the twenty-first century, race rarely took center stage in public discourse or policymaking.
While this does seem to be changing, particularly since the 2000's and the so-called “multi-cultural turn” (Rahier, 2018) that has seen many Latin American governments recognize their populations as multiracial and multiethnic rather than homogenously mixed, it remains the case that very few Latin American criminal justice departments are in the practice of systematically gathering (and fewer still, of publishing) ethnoracial data. This is why racial disparities research for incarceration in Latin America remains limited (but see Kalunta-Crumpton, 2012). Outside Brazil, in fact, the sum of published evidence for racial disparities across Latin American prisons (as far as we can gather) is as follows: One 2005 study conducted in the Colombian city of Cali found that the incarceration rate for Afrodescendent males was over 90% higher than that of non-Afrodescendent males (253/100,000 vs. 470/100,000, see Urrea-Giraldo, 2012). One 2000 study in Puerto Rico found that Black Puerto Ricans accounted for 8% of the general population but 12.9% of the prison population (U.S. Bureau of the Census, 2000). Incidentally, the year 2000 was the first time since 1950 that Puerto Rico's national census had collected racial data (Godreau et al., 2010). Aside from these two pieces of evidence, however, no published study outside Brazil has shared quantitative data for the ethnoracial makeup of Latin America's incarcerated populations, and none has compared the racial dynamics of incarceration across the Americas.
The Brazilian prison literature provides valuable insights for comparative research. In 2016, government estimates revealed that Black and Mixed race (Pardo) Brazilians comprised 64% of the prison population but only 53% of the general population (Departamento Penitenciário Nacional, 2016), indicating significant disproportionality, albeit less pronounced than in the US mainland. Scholars in Brazil have drawn parallels between the experiences of Afro-Brazilians and African Americans. João Costa Vargas (2011) characterizes policing in São Paulo as a “Black genocide” spanning the global African diaspora. Jamie Amparo Alves attributes police-perpetrated homicide in São Paulo to a “transnational neoliberal logic of black disposability” (2016: 230), one where Black Brazilians are conceived as the intended or “true” targets of police violence, and white populations as a kind of collateral damage (Alves, 2018: 53). These racial critiques of mass incarceration in Brazil are critically important in ending the silence on institutional racism in Latin America. The Black disposability paradigm in particular presents a powerful and necessary challenge to the infamous myth of “racial democracy”—now extensively poured over for over a century—that upholds Brazil in particular and Latin America in general as a racial paradise free from the structural racism that is seen by many to define the United States.
Despite these important contributions, however, we argue the Black disposability paradigm as applied to the problem of incarceration in Latin America also runs into some problems. One is that it has no theory of Mestizaje, even though mixing (as both demography and ideology) has been the single most significant feature of race formation in Latin America since at least the nineteenth century (Wade, 2005). A second problem with the Black disposability paradigm as applied to Latin America is that it ignores white imprisonment altogether, thereby presenting an incomplete account Latin American prison regimes (but see Denyer Willis, 2022).
Our study compares racial disparities in incarceration rates in Brazil, Cuba, Puerto Rico, and the United States mainland as measured by national census surveys. Our analysis finds that just as there are significant differences between Black, Mixed race, and white populations in their respective risk of incarceration, the degrees to which these categories correlate with the risk of incarceration also varies considerably internationally. The finding that profiles of racial inequality vary tremendously between countries does not minimize the importance of racism for understanding incarceration. Quite the contrary, it provides critical information that can inform our efforts to dismantle structural racism and mass incarceration as two incontestably intertwined structures. Before we describe these patterns, let us say a bit more about the census data we are using and our choice of countries.
Choice of cases and source
Any comparative study of racial inequality must reiterate that neither “race” nor the racial categories employed in population censuses are stable “things.” While rich and often highly generative sources of data, census categories are inherently complicated, owing to their lack of stability and changing meanings over time (Loveman and Muniz, 2007). This is especially true in Latin America, where census historians have noted the shifting meanings and boundaries of racial categories over the nineteenth and twentieth centuries (Loveman, 2009, 2014). Scholars routinely question the reliability of national census data on race, with some arguing that even censuses that include questions about ethnoracial identification can ‘hide’ the true extent of racial inequality (Bailey et al., 2013; Caraballo-Cueto and Godreau, 2021).
Partly in response to these challenges, scholars have proposed “colorism” and “pigmentocracy” as alternative concepts to replace ethnoracial categories of nation states (Dixon and Telles, 2017; Telles, 2014a, 2014b). These theories aim to measure skin color ‘objectively’ using scientific instruments. Methodologies include skin color palettes like the Von Luschan chart, reflectance spectrophotometry, and self-report scales (“light” to “dark”) (Caraballo-Cueto and Godreau, 2021; Swiatoniowski et al., 2013).
However, while colorism and pigmentocracy theories suggest a universal racial hierarchy, studies reveal significant variation in the correlation between skin color and inequality across Latin American countries. For instance, in Guatemala, individuals with “lighter” skin had double the education years compared to those with “darker” skin, while in Peru, the gap was smaller (Telles, 2014a, 2014b: 49). Similarly, in Colombia, Mestizo individuals had slightly more schooling than white individuals (Urrea-Giraldo, 2012: 109). These heterogenous findings not only complicate the notion of a universal “colorism” but they also provide useful signposts for advocates and policy makers working to advance racial justice.
For these reasons, we believe that using census data to comparatively analyse racial inequality is useful, though not because census data paint a perfectly complete or ‘objective’ picture of racial inequality; the contested nature of the construction of race means that no methodology can do this. Rather, census data provide a window into historically situated and internationally variable state constructions of race which can be used, in turn, to construct comparative portraits of racial inequality. As expressions of state power, censuses are particularly relevant tools for studying incarceration, which is among the most destructive examples and exercises of state power.
Regarding our selection of countries and territories, Brazil, Cuba, and Puerto Rico share similarities with the US mainland and with each other as former plantation societies with significant Afrodescendant populations. These populations, comparable in size to that of the US mainland, primarily emerged through the institution of slavery and the exploitation of enslaved African workers and their descendants on commercial plantations following European colonization. This history of plantation slavery distinguishes “Afro-America” (or “Plantation-America”) from “Indo-America,” where colonization relied more heavily on the exploitation of indigenous workers through forms of forced labor other than slavery (Best, 1968; Wagley, 1957). While studies on indigenous incarceration are necessary, particularly for Andean and Central American countries with large indigenous populations, we focused on Cuba, Brazil, the US mainland, and Puerto Rico due to their substantial Afro-descendant populations and the existing literature on anti-Black racism in areas beyond incarceration (De la Fuente, 2001; Gordon, 1949; Loveman et al., 2013; Telles, 2014a, 2014b). We excluded indigenous populations from our analysis because they constitute negligible proportions of the incarcerated population in these societies. Limitations of our country selection and sampling strategies are discussed further at the paper's end.
The data
The empirical basis for this study consists of census data from Brazil, Cuba, Puerto Rico, and the US mainland held in the International Public Use Microdata Survey (IPUMS) database. Our Brazilian data consists of a random 10% sample of the decennial census of 2010; our data for Cuba is a random 10% sample of the decennial census held in 2012. For both the mainland United States and Puerto Rico, for which demographic data are collected by the US Census Bureau, we have used the American Community Survey (ACS) and its Puerto Rican counterpart, the Puerto Rican Community Survey (PRCS), which are yearly surveys conducted on random 1% samples of the population. For the US mainland, we used the ACS survey from 2010 as this is the most recent available dataset. For Puerto Rico, given smaller sample sizes, we used the agglomerated sample for the five years between 2009–2013, creating a 5% total sample (the ACS and PRCS avoid sampling any one individual more than once in a given five-year period).
To garner information about our central variable— incarceration rates— we have used information recorded in all census surveys on “living arrangements,” focusing in particular on “group quarters,” which refers to living arrangements other than individual households. Within the broader category of “group quarters,” the Brazilian census specifies those who live in “prisons, reformatories or correctional institutions.” More challengingly, the surveys for Cuba, Puerto Rico and the US mainland, at least as coded in IPUMS, specify only those living in “institutional group quarters,” a larger category that in addition to prisons also includes (across the three countries): nursing homes, homes for the elderly, psychiatric institutions, homeless shelters, military or police barracks, student housing, and religious institutions. To try to discern the incarcerated population from this broader institutionalized population as closely as possible, we have added certain exclusionary criteria to remove populations living in other kinds of group accommodations. To exclude people living in homes for the elderly and student living quarters (the two largest categories of institutional group quarters other than prisons), we placed an age parameter of between twenty-one and sixty on all countries (including Brazil). However, since our data for Cuba, Puerto Rico and the United States mainland is not as precise as our data for Brazil, we have also sought to triangulate our figures, including both our extrapolated incarceration rate based on the “institutional” population and our estimates for the sex composition. The alternative sources we triangulate our estimates against include the World Prison Brief and various secondary sources. 1
Our Brazilian data sample is the only one that specifies incarcerated individuals, rather than simply individuals in institutional group quarters. It yields a total incarcerated population aged twenty-one to sixty of about 300,000 people, of whom 95% are male. Statistics from the World Prison Brief suggest an incarcerated population of just under 500,000 in 2010, of whom exactly 95% were male; it would make sense, given that we took only people aged twenty-one to sixty for the purposes of consistency with our Cuban, American and Puerto Rican data, that our estimated population size would be at least somewhat lower than the “real” one.
For Puerto Rico (see Table 1), our sample size of individuals in institutional group quarters is 1564. This translates into a population of 15,911, of which 88% were males. According to the World Prison Brief, Puerto Rico's incarcerated population was about 12,130 in 2008 and 12,244 in 2012, of whom about 96% were males. It would make sense that as our data covers the entire population in institutional group quarters between the ages of twenty-one and sixty (including psychiatric institutions, homeless shelters, religious institutions and so on), our figure would be larger than the actual number of incarcerated people. Given that we include these other institutions in our estimate, it also makes sense that the percentage of women in our data would be higher, since these other institutions are ordinarily not as heavily male dominated as prisons. While this is a working estimate of the size of the incarcerated population, this is the best approximation we can garner for the imprisoned population from available data.
Size and sex composition of incarcerated census samples and populations.
Our US mainland census data much more closely resembles that of the World Prison Brief. The “incarcerated” population deduced from our weighted sample of institutional group quarters inhabitants aged twenty-one to sixty is just short of 2.3 million, of whom 88% are male. As of 2010, there were 2,270,142 people in US prisons according to the World Prison Brief, of whom about 90% were male.
The Cuban census data we have garnered is by far the most uncertain, in part because official statistics on incarceration with which to compare are also highly contested. The population estimated from our sample is much lower, at less than 13,000, than the official Cuban government figure from 2012 cited in the World Prison Brief: 57,337. Additionally, the gender ratio in our sample is very suspicious. At 62% male, according to our estimate, this implies a much higher percentage of incarcerated women than is typical of prisons in most countries. Unfortunately, this gender breakdown is impossible to triangulate. The World Prison Brief does not specify the sex ratio of the Cuban incarcerated population, as it does for most countries (and neither does any other database, as far as we know), nor is there readily available data on this question. Therefore, in either case, it seems clear that our Cuban data should be by far taken with the greatest grain of salt.
Before we proceed to the data on racial disparities in incarceration, a note on the racial identification categories by which these data are grouped. For all four countries/ territories, we focus our analysis on three main ethnoracial categories—a decision which is convenient, but also grounded in the realities of census results. Each of the four sites (the United States mainland, Puerto Rico, Cuba, and Brazil) have racial categories called “white” and “Black,” and in every site more than half of the population subscribed to one of these two categories in censuses around the year 2010. In Cuba, the entirety of the population, and in Brazil, 98%, identified with one of three racial categories— white, Black, and either “Mixed race” (Mulato, Cuba) or “Brown” (Pardo, Brazil).
Racial categories in Puerto Rico and the United States, which both use US Census categories, are somewhat more complicated. In the Census of 2010, 70% of Puerto Rico's population identified as “white,” and another 10% as “Black.” In addition, roughly 10% each identified either with more than one race, or simply as “other”— a reflection of how US census categories do not quite fit in a society where the Black-white binary has never captured the local spectrum of racial identification (Godreau et al., 2010). In our analysis of Puerto Rico, we combine the “Other” and “Two or more races” categories to create a third category (“Mixed race”) which, when combined with Black and white, covers virtually the entirety of Puerto Rico's population.
Finally, the United States census categories are perhaps even more complicated, because of the treatment of a separate “Hispanic” ethnicity, separate from racial identification (see Table 2). In Puerto Rico, virtually the entire population identifies as ethnically “Hispanic” and therefore it cannot be meaningfully treated as a separate category. On the United States mainland, on the other hand, the label “Hispanic” tends to be treated as a distinct ethnicity which may be elected in addition to racial categories (usually Black and white). Moreover, Hispanics are also often analysed as a distinct group in studies of imprisonment, with Hispanic individuals (of all races) identified along with non-Hispanic Black people as a major group overrepresented in American prisons (U.S. Bureau of the Census, 2000). In measuring racial inequality, then, we have decided to look at three major groups in the United States: non-Hispanic Black individuals, non-Hispanic white individuals, and Hispanics (of all races). Thus, our ethnoracial categories for the US mainland are distinct to those of the other three sites in that the third category is “Hispanic” rather than “Mixed race.” Together, non-Hispanic Black individuals, non-Hispanic white individuals, and Hispanics (of all races) constituted 92% of the US population aged twenty-one to sixty in 2010 and 95% of the imprisoned population aged twenty-one to sixty.
Ethnoracial categories from national censuses.
*US mainland also includes “Hispanic” as an “ethnicity,” to be chosen in addition to the above “racial” categories
Racial disparities in incarceration rates
To analyze racial inequality in incarceration rates, we use a straightforward measure: a disproportionality index. This index calculates each ethnoracial group's representation in prisons compared to their representation in the general population as a percentage. Our analysis focuses on individuals aged twenty-one to sixty to better reflect the incarcerated population within the total institutionalized population. Table 3 illustrates this pattern for both males and females combined.
Racial disproportionality incarceration*.
*“Racial disproportionality index” for each ethnoracial group = (share of incarcerated sample/share of general population) × 100.
Now to the inequalities themselves. Three clear commonalities emerge: white populations are underrepresented in prisons, while Black populations are overrepresented. In Puerto Rico, Brazil, and Cuba, people who identify as Mixed race and over-represented in prison, and on the US mainland, people who identify as Hispanic are over-represented in prison. These clear commonalities indicate that racism pervades criminal justice systems beyond the US mainland. However, the extent of racial inequality varies among the societies studied. The US mainland and Cuba demonstrate relatively high racial inequality, while Brazil and Puerto Rico exhibit comparatively lower levels. This is evident when examining the underrepresentation of white demographics. As shown in Table 3, in Cuba and the US mainland, white individuals are incarcerated at 59% and 61% of the expected rate, respectively, whereas in Brazil and Puerto Rico, they are incarcerated at 79% and 72% of the expected rate.
Let us consider in more depth each site in turn. The United States mainland stands out in the extent to which Black populations—in this case, non-Hispanic Black people—are overrepresented among the incarcerated. Here, the Black demographic share of the institutionalized population aged twenty-one to sixty, including both males and females, is nearly three times their share of the general population. Hispanics (of all races) are also overrepresented in the imprisoned population, but by a much lower factor of 1.17.
Puerto Rico displays lower racial inequality than the US mainland, with Black individuals overrepresented by 72%. This concords with comparative data compiled by the Prison Policy Initiative, which found that while Black populations were overrepresented by a factor of 3.5 in the US, the figure was 1.5 in Puerto Rico (Prison Policy Initiative, 2004). Mixed race Puerto Ricans (those that identify as “Two or more races” or “Some other race”) are also over-represented in prison, at 154% their population share, a significant disproportionality but a less sharp one relative to Black Puerto Ricans.
In Brazil, the percentage of incarcerated who are white comes closest of all four sites to the share of white people in the general population—but still only to 79%. For their part, Black Brazilians are somewhat less overrepresented in prisons than in Puerto Rico, at 166% versus 172%, with Pardos also overrepresented in prison given their share of the population but to a lesser degree than Black Brazilians, at 107%. These figures concord with figures from the Brazilian government suggesting that white Brazilians made up 37.22% of the country's imprisoned population versus 45.48% of the general population, while the figures for Black individuals and Pardos combined (the two groups were not disaggregated) were 61.67% and 53.63%, respectively (Departamento Penitenciário Nacional, 2014: 36). Notably, the decision to treat Black people and Pardos as a collective in the government estimates—something many race scholars and activists advocate (Reis and Crespo, 2015; Santos, 1999)— in the case of incarceration seems to present a flattened out portrait of racial inequality; partly because the number of Brazilians who identify as Pardo is much greater than the number who identify as Black. When we look at these two groups separately—as we do in our analysis—Black populations are clearly disproportionately incarcerated to a much greater degree than when they are combined with Pardos.
Cuba presents the most stark racial inequality, with white Cubans underrepresented at 59%, exceeding the underrepresentation of white individuals on the US mainland (at 61% their population share). The overrepresentation of Black individuals is remarkably high in Cuba, albeit not as high as in the US (277% vs. 297%), with the figure for Mixed race or Mulato/a individuals also elevated (136%) but not as starkly as for Black Cubans (277%). However, caution is warranted with Cuba's data since our estimated population of “institutionalized” twenty-one- to sixty-year-olds is significantly lower than other (limited) sources on the country's incarcerated population (World Prison Brief, 2022).
Domains of racial inequality: an initial comparative discussion
One clear pattern seems to emerge from the figures discussed in the previous section: the United States mainland and Cuba are characterized by relatively high racial inequality in incarceration rates, while Brazil and Puerto Rico exhibit comparably lower inequality. What might explain this pattern? The principal objective of this study is to provide a preliminary picture of varying patterns of racial disparity in incarceration in Latin America. Our data do not yet allow for conclusive causal explanations for these variations. Nevertheless, we would like to provide several preliminary interpretations that might help set the agenda for future comparative research into racial inequality across the Americas.
A robust literature from the US mainland links the evolution of mass incarceration and huge racial disparities in incarceration to historical patterns of political economy (Muller and Schrage, 2021). Moreover, on the US mainland these large racial disparities in incarceration coexist with large socioeconomic inequalities along racial lines. One first step in explaining the varying patterns of incarceration disparities we have examined is to examine whether these disparities correlate with socioeconomic inequalities.
Table 4 presents findings from another comparative study of racial inequality in Latin America (Ayala-McCormick, 2021). The data summarized consist of two additional metrics of inequality: the median earned income of employed males, and the rates of college education in the general working-age (twenty-five to sixty-four) males on the other. In both cases, the figures for Black and Mixed race males are displayed as a percentage of the figure for white males. These data present something of a paradox when compared to our data on incarceration. In particular, Cuba and Brazil switch places. That is, in terms of income and education, Brazil and the United States are characterized by higher inequality, and Cuba and Puerto Rico by lower inequality. Thus, for example, the median income among Black and Mixed race men in Brazil was 74% of the figure for white men— in the US, the median Black male made two-thirds what the median white male made. In Puerto Rico, in contrast, the figures for Black and Mixed race men were 90% and 95%, respectively. A similar pattern emerges for university education: the percentage of Black and Mixed race men with a university degree was less than a third of the rate for white men in Brazil, and one-half in the United States; in Cuba and Puerto Rico, on the other hand, the figures were much higher.
Measures of socioeconomic inequality (males only, black and mixed race figures as % of white figure).
Source: Ayala-McCormick (2021). Figures are for 2010 in Brazil, 2012 in Cuba, and 2018 in Puerto Rico and the US.
Table 5 above thus considers the intersection of these two domains of racial inequality: disparities in incarceration rate and socioeconomic inequality. On one extreme, the US mainland displays higher levels of inequality in both domains. On the other extreme, Puerto Rico seems to have lower levels of inequality in both domains. In the cases of Brazil and Cuba, however, the two domains do not mirror each other: higher socioeconomic inequality coexists with lower incarceration disparity in Brazil, while lower socioeconomic inequality coexists with higher incarceration disparity in Cuba.
Socioeconomic inequality vs. incarceration disparity.
To try to explain variations in racial disparities in incarceration in terms of their correlation with racialized economic inequality, one important puzzle concerns the paradoxical pattern in Brazil and Cuba. We examine several possibilities to guide future research below.
Racial geographies and racial inequalities
Critical research is needed to better understand the relatively lower racial inequality seen in Brazilian prisons relative to the country's other socioeconomic indicators including wealth, income, and education (see Ayala-McCormick, 2021). One intriguing line of inquiry can be found in Latin American studies of racial geography. Since the mid twentieth century, scholars have noted that Latin American ethnoracial identities can ‘vary’ in line with changes in geography and socioeconomic status (Degler, 1971; Mintz, 1956), an observation that finds support in recent studies. Puerto Ricans who live on the island and who identify as “white” in the Puerto Rican census have been known to change their racial identity to “Black” upon emigrating to the US mainland (Duany, 2005). Furthermore, studies of racial geography in Brazil and Puerto Rico have shown how certain spaces—low-income neighborhoods and heavily policed public housing projects—become culturally coded as Black (Dinzey-Flores, 2013; Rio Caldeira, 2000; Rivera-Rideau, 2013). Building on racial geography's argument that spaces may be constitutive of “race,” more studies are needed to explore how imprisonment may impact racial identification. In particular, such a study in Brazil may help decipher the Brazilian paradox we noted above: Could it be that imprisonment changes how Brazilians identify racially within the census? If so, could this interaction between imprisonment and racial identification help to explain Brazil's surprisingly lower rate of racial inequality in incarceration relative to its high levels of racial inequality across wealth and income?
Historical racial regimes
A second conceptual tool that may be useful for comparing racial inequality internationally is that of “historical regime,” operationalized recently by sociologist Regina Baker in order to analyze variations in the difference in poverty rate between various states in the US South. The key independent variable developed by Baker to explain these variations, labelled “historical racial regime,” was a compound of the percentage of the state's population that was enslaved in 1860, the percentage of sharecroppers who were Black in 1930, as well as various measures of overt racist or segregationist policy in the Jim Crow era (Baker, 2022). Historical racial regimes may be helpful for shedding light on some of the paradoxes identified in this paper.
Compared to Puerto Rico or Brazil, Cuba historically had a higher percentage of enslaved people and a lower percentage of free people of color within its total population. In Cuba in 1800, for example, 35% of the population was enslaved while 19% were free people of color; in Puerto Rico, the figures were starkly reversed, at 15% and 40%, respectively (Andrews, 2016: 41). In 1872 (the first time Brazil conducted a nationwide census), 15% of the population was enslaved, while 42% of the population were free people of color (O’Neill, 2022). Could these differences in “historical racial regime” have set the pattern of racial inequality and incarceration that we see across the four sites? That is, do historical differences in the respective population share of enslaved versus free people of color help to explain different levels of incarceration and racial inequality seen today? More research comparing distinct racial regimes historically across Latin American nations might shed light on the distinct profiles of racial inequality identified in this paper.
The clearest correlation: racial disparity and absolute incarceration rate
Up until now, our interpretation of the pattern of racial disparities in incarceration between the four cases has been suggestive rather than conclusive. We should note that there is one clear correlation that might help explain the pattern of racial disparities we have observed: that between the magnitude of racial disparities on the one hand, and the absolute incarceration rate on the other.
Table 6 shows estimates of the absolute incarceration rate (number of incarcerated people per 100,000 inhabitants) derived from the World Prison Brief for each of the four societies in 2002 and 2012. If we take our Cuban data as a very rough measure of racial inequality in incarceration, and if we elect to use World Prison Brief data instead to estimate the absolute size of Cuba's prison population, a clear pattern emerges. Societies with higher absolute incarceration rates (the United States mainland and Cuba, at 510 and 707, respectively) have much sharper racial disparities than societies with lower absolute incarceration rates (Brazil and Puerto Rico, at 270 and 339, respectively). This pattern—whereby racial disparities increase as incarceration rates rise—would seem to suggest that as incarceration rates increase within a society, the protective benefits conferred by whiteness also increase. This interesting yet understudied characteristic of racial inequality—its tendency to be relational (Riley, 2020)—merits further investigation. Were this pattern to hold true for other countries beyond our analysis, this would indicate that reducing the absolute incarceration rate is not only good in itself but it might also have a secondary positive effect in reducing racial disparities.
Incarceration rates in 2002 and 2012 (prisoners per 100,000 inhabitants).
Source: World Prison Brief.
Conclusion
This study reveals that prisons across Brazil, Cuba, Puerto Rico, and the US mainland exhibit consistent patterns of racial inequality, with Black populations overrepresented and white populations underrepresented. Mixed-race individuals are also overrepresented in Latin American nations, while Hispanics experience overrepresentation on the US mainland. These findings challenge the notion that institutional racism within criminal justice is a peculiar quirk of the US mainland (Bourdieu and Wacquant, 1999) and provide important empirical support to ongoing data collection efforts to document institutional racism in Latin America (Schaumberg et al., 2021).
Our second key finding is that racial inequality varies considerably internationally. While race and ethnicity scholars have long acknowledged that there are “multiple racisms” (Gilroy, 1991), scholars have a harder time contending with the notion of stronger or milder manifestation of racism. We think here of Hilda Llorén's (2020) recent reflections on racism in Puerto Rico, in which she stated that “there is no such thing as a ‘less violent’ form of anti-Black racism.” While we acknowledge the historical context of racism-denialism in which statements such as these are made, our findings lead us to a different conclusion that we think is important for the purposes of advancing racial justice. There are absolutely ‘more’ and ‘less’ violent forms of racism. In the present example, these different levels of antiblack racism manifest as internationally variable levels of racial inequality. Put simply: if you are Black in Puerto Rico you are significantly less likely to be incarcerated than if you are Black in the United States mainland. Racism is clearly not monolithic.
Our third finding is that the racialization of incarceration applies not just to Black and Mixed-race people but to white people too. Stated plainly, white people's risk of imprisonment is reduced by racism within white supremacist societies. Our finding that whites are consistently under-represented in prison internationally highlights the fact that white imprisonment is not a simple benchmark against which to compare other races.
Moving to our fourth observation, we stress the need for research on incarceration and racism that grapples with the full ethnoracial breadth of the prison population. The problem with focusing exclusively on Black imprisonment becomes clear when we consider the moves authors have sometimes had to make in order to account for white imprisonment (we will come to the problem of Mixed-race imprisonment shortly). Scholars writing in the Black disposability paradigm tend to employ some version of the ‘collateral damage’ thesis. In the context of the US mainland, for example, Michelle Alexander states this most plainly when she says: “the fact that white people are harmed by the drug war does not mean they are the real targets” (Alexander, 2012: 205). Using the analogy of Iraqi civilian death in the war against terrorism (an analogy we suspect could do with unpacking, see Perugini and Gordon, 2017), she states that “Black and brown people are the principal targets in this war; white people are collateral damage” (Alexander, 2012: 205). Making a similar argument, but this time in the context of Brazil, Alves (2018: 53) characterizes the murder of white Brazilians by police as “an accident” ascribable to whites’ “proximity to the black body.” While we acknowledge there are important theoretical distinctions among these authors, we warn that the collateral damage hypothesis falls short when it comes to understanding the factors driving the recent explosion of incarceration across Latin America. This becomes clear when we look at the regional scale of incarceration today.
While our study supports the notion that white people are protected from incarceration due to their race across the Americas, it's crucial to recognize that white incarceration is far too high (in absolute terms) across all of the countries studied to provide a useful target for those of us seeking to dismantle structural racism. The fact that over 1.5 million people are incarcerated in the Latin American region (World Prison Brief, 2021)—and that in this study, over half (53%) of Puerto Rico's prisoners were white, with white prisoners accounting for 40% of the prison population in Brazil, 37% in Cuba, and 38% on the US mainland—is a powerful illustration of the fact that when it comes to being incarcerated, white privilege is very easily overridden. So rather than reading incarceration as a lousy or imprecise tool for achieving its ‘primary’ goal (enforcing antiblack racism), the more accurate reading is that there are multiple logics driving up incarceration. By far the best understood of these are market logics that penalize poverty (Sozzo, 2016).
The challenge for future comparative research is that of teasing apart how antiblack racism and the penalization of poverty might differentially articulate with each other internationally. This is a line of inquiry that will be greatly impoverished if researchers focus only on Black incarceration. Just as gender theorists cannot begin to understand how gender shores up differential life opportunities to men and women by focusing only on women, racial critiques of mass incarceration in Latin America cannot adequately grasp how racism shapes incarceration by focusing only on Black imprisonment.
This brings us to our fifth and final observation. For any racial critique of incarceration to have local relevance in Latin America, it must have something to say about racial mixing. For at least two centuries, after all, Mestizaje has arguably constituted the single most significant demographic and ideological process shaping race formation in Latin America (Miller, 2004; Wade, 2005). When scholars choose to not distinguish between Black and Mixed race imprisonment, or when scholars recode Latin Americans who identify as “Pardo,” “Mulato,” or “Two or more races” as Black—they elide stark differences in the respective risk of incarceration experienced by Black and Mixed race people. This is especially clear in the cases of Cuba and Brazil. By our estimates, Black people in Brazil are disproportionately incarcerated at a rate that is 66% higher than their share of the total population, while Pardos are incarcerated at a rate that is just 7% higher than their demographic representation in the general population (see Table 3). Similarly, Black people in Cuba are incarcerated at a rate that is 177% higher than their share of the total population, while Mulatos are incarcerated at a rate that is 36% higher. When scholars treat Black and Mixed race people as one single category, the overall effect can be to underestimate the true extent of racial inequality experienced by Black Brazilians and Black Cubans. Yet such differences in respective risk of incarceration between Black and Mixed race people matter. For perspective, the elevated risk of incarceration experienced by Mulatos in Cuba and Hispanics on the US mainland vastly outstrips the elevated risk of incarceration experienced by Blacks in Puerto Rico and Blacks in Brazil (see Table 3).
All this striking variation between the experiences of Black and Mixed race populations on the one hand, and the different ways this plays out internationally on the other, merit our attention. Future research might consider, for example, how racial identification is differently classed internationally. Could it be, for example, that Black identification in Puerto Rico has a different class profile to Black identification in other places? Census analyses would be greatly strengthened by richer understandings of how whiteness and blackness are conceptualized by survey respondents and of how these understandings may differ internationally and geographically within nations as well as over time.
There are several limitations to the present study that point to future lines of research. First and foremost, our internationally comparative census data stored in international IPUMS database pertained directly to incarceration only in Brazil. For Puerto Rico, Cuba, and the US mainland, the racial breakdown of the incarcerated population was inferred from the census count of the “institutionalized” population using techniques of triangulation. We hope that more granular internationally comparable datasets that will become available in years to come, which would greatly improve the robustness of these preliminary findings. Additionally, while the present study is useful as a tool for painting a preliminary portrait of current quantitative differences in profiles of racial disparities, more detailed comparative historical studies are needed to comprehensively grasp how national histories of slavery, forced labor and distinct racial regimes may shape current ethnoracial disparities at the level of incarceration.
Finally, our study has nothing to say about indigeneity. In Cuba and Brazil, there is no census category for indigenous people, which itself could be contributing to the erasure of indigenous groups in both countries. Both the US mainland and Puerto Rico do enumerate the indigenous population, but our analytic approach of selecting random samples of the population may have obscured this smaller population. Given the small percentage of indigenous representation in our case study countries compared to other countries in the region, our comparison of racial inequities in incarceration could well yield very different results if undertaken instead among Andean or Central American countries or those of the Southern cone. Finally, this study utilizes the most recent data available for Brazil and Cuba from the international IPUMS database which is now over a decade old. Efforts are needed to boost access to more recent internationally comparable census data.
It is too early to tell what future comparative research into incarceration's ethnoracial dynamics will reveal. But this is undoubtedly a conversation those of us who seek to dismantle incarceration and structural racism must participate in, if we are to bring the disastrous era of mass incarceration to an end.
Footnotes
Acknowledgements
The authors gratefully acknowledge Diego Alaya-McCormick, who contributed to an early draft of this paper, as well as research assistance from Dorothy Apedaile, Samantha Murphy, and Jen Viken.
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by The British Academy/Leverhulme Trust (grant number SRG21\210718). Dr Parker was PI for this grant.