The legacies of slavery in and out of Africa

4.2 The USA

Slavery was introduced in the territories that today represent the USA in the sixteenth century, much later than in Spanish South America and Brazil. The scope was to replace European and African indentured servants as the main source of plantation labor, at the time mostly employed for the cultivation of rice and tobacco. Between 1675 and 1695, the import expanded rapidly. By the 1720s, Virginia and Maryland had been transformed into slave societies. Overall, the inflow to the USA, throughout the next centuries, amounted to an estimated 645,000 slaves, brought in mostly from Africa. The slaves were initially disembarked along the Atlantic coast and forcibly settled in the coastal Southern colonies. Even though the USA absorbed less than 4 % of the entire volume of the trans-Atlantic trade, the local reproduction rate was much higher than elsewhere so that the slave population, unlike in the rest of the Americas, expanded. By the 1730s, births to slave women outnumbered import, with an increase of the African population at an annual rate of 3 %. As a result, at the start of the American Revolution, the region was no longer an immigrant society. Later on, in the 1789–1860 period between the Revolution and the Civil War, most slaves were relocated in the inland regions where the plantation economy was quickly expanding following the booming international demand for cotton. This Second Middle Passage ended only with the Confederate defeat in the Civil War. Despite the fact that the Revolution broke the coincidence between blackness and slavery, between 1800 and 1860 the slave population increased from one to four million, so that by the 1860 census the USA had a slave population of about 13 % of the total, distributed within 15 slave states, mostly belonging to the South. The American Civil War led to the abolition of slavery in 1865. The Reconstruction period, running from 1865 to 1877, witnessed a transformation of Southern society and the enactment of legislation favoring the rights of former slaves. However, soon afterwards the white elites were able to restore their control and to introduce restrictive Black Codes and disfranchisement provisions. The next massive movement of the African American population occurred between 1916 and 1930, with the so-called Great Migration from the rural South to the urban North, pulled by new job opportunities in the Northern cities and pushed by the crisis of the cotton economy. The latter was caused by the boll weevil beetle infestation and also by the social and political conditions of blacks in the South. Black emigration from the South slowed down after 1930 but picked up again after World War Two. It continued at differentiated speeds until the 1970s, reaching a total volume of six million, with a partial reversal after that.

Early contributions on the economic impact of slavery in the USA include the influential although controversial book by the economic historians Fogel and Engerman (1974), where they argue that slavery in the antebellum South was an efficient production arrangement. Contrasting views were expressed among others by David and Stampp (1976) and Ransom and Sutch (2001). The more recent literature I focus on has investigated the long-term consequences of slavery on development. Across states over the 1880–1980 period, Mitchener and McLean (2003) find a negative and persistent effect on productivity levels. Lagerlöf (2005) explores the link between geography and slavery and also uncovers a negative relationship between slavery and current income. Both across states and counties, Nunn (2008b) reports a negative effect of slavery on per capita income in 2000.

Using the dataset collected by Bertocchi and Dimico (2010), in Table 4 I provide empirical evidence on the cross-county influence of slavery on the contemporaneous level of development in the USA. Slavery is measured as the share of slaves to the total population in 1860, while the dependent variable is per capita income in different years. After entering geographical controls meant to capture structural differences among different regions of the USA (i.e., dummies for counties within former slave states and for counties within North-Eastern and South-Atlantic states), the relationship is not significant for per capita income in 2000. Across previous decades, the relationship was still significant in 1970, but no longer so in 1980 and 1990. This suggests that the effect of slavery on income is not a robust one.

Turning to the link between slavery and current inequality, in Table 5 I present results for different indicators, all measured in year 2000: income inequality and racial inequality (both computed as Gini indexes) and the fraction of the population below the poverty level. Using the same specification as in Table 4, i.e., controlling for structural differences across regions, for all dependent variables, slavery always retains a positive and significant coefficient. Thus, there is robust evidence that the distribution of per capita income is more unequal today in counties associated in the past with a larger proportion of slaves in the population, and so is the racial dimension of inequality, while poverty is more widespread.

Furthermore, over a state-level panel dataset of educational attainment across races over the 1940–2000 period collected by Bertocchi and Dimico (2010), in Table 6 I regress the educational racial gap, at the high-school and bachelor level, on the share of slaves in the population in 1860: the coefficient is significantly positive, which suggests that the impact of slavery may run through the evolution of the educational gap (Table 6). Indeed, after the Civil War and abolition, illiteracy was predominant among blacks and progress was very slow until the eve of World War Two.

The hypothesis according to which human capital formation may represent the channel through which the effect of slavery still lingers on in American society echoes of a large literature on race and human capital including Smith (1984), Margo (1990), Sacerdote (2005), and Canaday and Tamura (2009). The same hypothesis is further developed and tested in Bertocchi and Dimico (2014), where my coauthor and I employ a Theil decomposition to disentangle the two components of income inequality: inequality across races (racial inequality) and inequality within races (within inequality). The negative and significant influence of slavery is confirmed after controlling for factor endowments and running two-stage least-square regressions. An alternative hypothesis might attribute the effect of slavery on current inequality to racial discrimination. Indeed the bond between slavery and racism, which was non associated with slavery in the Old World and is much weaker in today’s Latin America, is perceived as particularly strong in the USA. To test this additional hypothesis, we create a measure of racial discrimination based on returns to skills, estimate returns to education for blacks and whites, and compute the ratio of average returns for blacks to average returns for whites. The latter turns out to be well below 1, consistently with the presence of discrimination. Using this proxy, we do find that racial discrimination contributes to inequality, but to a much lesser degree if compared to the human capital transmission channel. This conclusion is consistent with Fryer (2011), who argues that relative to the twentieth century the relevance of discrimination as an explanation for racial inequalities has declined, since racial differences are greatly reduced when one accounts for educational achievement. We conclude with suggestive evidence that the underlying links between past slavery and current inequality run through the political exclusion of former slaves and the resulting negative influence on the local provision of education for black children.

In a companion investigation also by Bertocchi and Dimico (2012a), my coauthor and I shed further light on the evolution of racial educational inequality across states from 1940 to 2000, extending the results illustrated in Table 6. Despite a gradual reduction of the gap over this period, the evidence shows that the racial gap at the high-school and bachelor level is determined by the initial gap in 1940, which is in turn largely explained by past slavery. The correlation between the racial educational gap in 1940 and the share of slaves over population in 1860 is in fact 0.90 and 0.81, at the high-school and bachelor level, respectively. Two-stage least-square regressions where slavery is used as an instrument for the initial gap confirm this conclusion. The issue of the excludability of slavery is addressed by instrumenting it with the share of disembarked slaves from the trans-Atlantic slave trade, i.e., accounting for the link between the geographic slave distribution following the Middle Passage and the one prevailing after the Second Middle Passage. We also find that income growth over the same period is negatively correlated with the initial racial gap in education, which suggests that slavery also exerts an indirect effect on growth through the education channel.

In Bertocchi and Dimico (2012b), my coauthor and I extend the analysis of the political implications of slavery using a unique dataset on voting registration by race assembled for the counties in the state of Mississippi in 1896, in the middle of the period that witnesses the restoration of the white elites’ supremacy. We show that the disfranchisement measures introduced with the new 1890 state constitution (i.e., the requirement of a poll tax and a literacy test for voting registration) negatively affect the political participation of blacks. However, we also show that the decline starts even earlier, reflecting a process of institutionalization of de facto disfranchisement and thus supporting the fait accompli hypothesis advanced by Key (1949). Black registration is shown to be more limited in the presence of a larger share of black population, which is in turn highly correlated with a larger share of slaves before abolition. This can be explained by the fact that a majority of black voters represents a more serious threat to white supremacy. The paper also shows that restrictions in black political participation affect educational policies in a persistent fashion, consistently with the previously mentioned contributions. Naidu (2012) also contains an analysis of the consequences of the disfranchisement measures introduced in Southern states on political and educational outcomes. Acharya et al. (2016) show that contemporary differences in political attitudes still reflect the intensity of slavery in 1860, with Southern whites more likely to support the Republican party and oppose affirmative action policies in counties more affected by slavery historically. They interpret these results as the long-term consequences of the conservative political attitudes that developed after the Civil War.

Again with special attention to Mississippi data, Chay and Munshi (2013) focus on the subsequent epoch which, between 1916 and 1930, witnessed the Great Migration of one million of former slaves from the South to the North of the USA. They find that blacks coming from counties characterized by labor-intensive plantation crops represented a disproportionate share of Northern migrants. They attribute this finding to the development of social network externalities that became instrumental in the mobilization process when large coalition of blacks moved together to Northern cities.

As for the case of Africa, the influence of slavery on gender roles and cultural norms has been investigated also for the case of the USA. Mohinyan (1965) argues that the structure of the black family has been undermined by slavery, with broad consequences on crime and the social condition of blacks. Slavery has also been proposed as an explanation for the racial gap in female labor force participation. Boustan and Collins (2014) show that for over a century, that is from 1870 until at least 1980, black women were more likely than white women to participate in the labor force and to hold jobs in agriculture or manufacturing. They also show that differences in observables cannot fully account for this racial gap, which confirms the intuition in Goldin (1977). The latter suggests that female labor force participation reflects a “double legacy” of slavery. A direct effect may have come from the low levels of income and education for blacks, which pushed more black women into the labor market. Moreover, an indirect effect may have come from an intergenerational transmission channel: since black women were forced to work intensively under slavery, African Americans developed different cultural norms about women’s work, with consequent long-term effects. Another cultural implication of slavery is examined by Gouda (2013), who shows that the slave share in 1860 is correlated with contemporary violent crime, suggesting that the culture of violence that developed under slavery still exerts a lasting effect.

Beside education, human capital is also shaped by health conditions. The hypothesis that the racial gap in life expectancy may be linked to the slave trades has been advanced by Cutler et al. (2005), who present evidence suggesting that racial differences in sensitivity to salt, a leading and largely hereditary cause of hypertension, may be due to selection during the Middle Passage. Because of intense water loss, the ability to retain salt and hence water substantially increased the chances of survival, which induced slave traders to select captives on the basis of the salt on their skin. Bhalotra and Venkataramani (2012) find that the impact on adult education and labor market outcomes of the reduction in pneumonia in infancy—thanks to the introduction of antibiotic therapies in the 1930s—decreases with the intensity of slavery in 1860. They interpret this result as a consequence of pre-Civil Rights barriers to realizing returns to human capital investment for blacks born in the South. The genetic resistance to malaria of African slaves has been suggested by Mann (2011) as the reason why slavery developed in the USA, and indeed Esposito (2013) documents a correlation between malaria suitability and the diffusion of slavery as well as slave owners’ preferences for slaves more likely to be immune.

To conclude, the evidence for the USA points to a robust influence of past slavery on inequality, while the influence on current income levels is somewhat weaker. The main channel of transmission is to be found in the unequal access to education and the accumulation of human capital for the descendants of slaves. The political mechanism behind the local provision of funding for schools determined an inferior level, both qualitatively and quantitatively, for the education inputs accessible for black children, with lingering consequences to the present day.

Despite the fact that the Civil Rights movement and legislation have removed the most visible vestiges of slavery half a century ago, the debate of the consequences of slavery in the USA is still open. Julian L. Simon influentially contributed to it by proposing a calculation of the black reparations bill, which he estimated to amount to about $58 billion, i.e., about 7 % of annual GDP (Simon 1971). The awareness that the lingering influence of the history of blacks in America runs through the human capital channel is witnessed by the declared goals of recent federal education programs, from Bush’s No Child Left Behind to Obama’s Race to the Top, which have aimed at the removal of the racial and ethnic educational gaps that persistently afflict the American society. At the same time, however, the fact that in the USA inequality displays a strong racial component has not been sufficiently emphasized in the recent debate on the long-term evolution of income and wealth inequality, propelled by the book by Piketty (2014). Indeed in his analysis of inequality, Piketty (2014) only very briefly mentions the racial gap in wealth, despite the fact that—as reported by The Economist (2015)—the median white family in 2013 owned net assets almost thirteen times larger than the median black family. Likewise, Putnam (2015) stresses the widening divide in attitudes toward nurturing children within all racial groups, thus shifting the focus from race to class as a driver of differences in educational achievement. He claims that achievement gaps between rich and poor pupils belonging to the same race are now larger than those between races of the same income level. In other words, according to his analysis, the class gap has been growing within each racial group, while the gaps between racial groups have been narrowing. However, his conclusions can be challenged on the ground that they are driven more by the worsening performance of poor whites rather than by the improvement of that of blacks.