The racial gap in education and the legacy of slavery

https://doi.org/10.1016/j.jce.2012.04.001Get rights and content

Abstract

We study the evolution of racial educational inequality across US states from 1940 to 2000. We show that throughout this period, despite evidence of convergence, the racial gap in attainment between blacks and whites has been persistently determined by the initial gap. We obtain these results with 2SLS estimates where slavery is used as an instrument for the initial gap. We address the question of the excludability of slavery by instrumenting it with the share of disembarked slaves from the Trans-Atlantic Slave Trade. Using the same approach we also find that income growth is negatively affected by the initial racial gap in education and that slavery affects growth indirectly through this channel.

Introduction

Among the several dimensions of inequality, racial inequality is a rather peculiar characteristic of American society, if compared to other countries at similar levels of development. Because of the crucial role played by human capital in determining earnings prospects, the unequal distribution of income between blacks and whites is in turn closely associated with the underlying racial gap in educational attainment. Over time, black–white gaps in schooling have actually decreased.1 At the same time, however, this trend has come to a halt in more recent decades,2 confirming the persistence of deeply-rooted racial disparities despite the observed process of convergence. These recent developments suggest that the dynamics of the racial gap in attainment follow path dependence, i.e., they are influenced by initial conditions.

To test the hypothesis that racial educational inequality persists over time and depends on its initial level, which can be measured from the year 1940,3 would inevitably expose the empirical analysis to omitted variables and measurement errors bias leading to inconsistent estimates. To address this problem, it is legitimate to turn the attention to the association between the racial gap in education in 1940 and past slavery. Collins and Margo (2006) narrate how in the aftermath of the Civil War the “legacies of slavery” determined extremely high rates of illiteracy among blacks. Margo (1990) documents how the obstacles subsequently encountered by black children in acquiring education represented the channels through which past slavery ended in inequalities which were still present at the eve of World War Two. Indeed the correlation between the racial educational gap in 1940 and the share of slaves over population in 1860 is extremely high: 0.90 and 0.81 at the high-school and bachelor level, respectively. However, to use past slavery as an instrument for the initial gap in education raises questions regarding its excludability, which may not be warranted in the presence of any direct effect on educational outcomes.

The goal of this paper is twofold. First, we aim at obtaining empirical estimates of the persistency of the effect of the initial gap on subsequent educational disparities between whites and blacks. Second, we try to establish that past slavery – being a crucial determinant of the initial gap – still plays a considerable role in American society.

We can present our empirical strategy and our results in two steps. First, we look at the determinants of the racial educational gap over the 1940–2000 period using a simple OLS estimator, with a focus on the potential effect of past slavery. This preliminary step reveals that slavery in 1860 does appear to influence the dependent variable only through the initial gap. This indirect effect of slavery therefore suggests that slavery may be a suitable instrument for the latter. However, omitted variables and measurement errors bias may once again be present in the relationship between slavery and the initial gap, yielding inconsistency of OLS and consequent under-estimation of the effect of the initial gap on subsequent inequality. Moreover, the excludability of slavery needs to be assured more formally. To address these problems, we use an instrument for slavery in 1860 which we construct on the basis of information on the Trans-Atlantic Slave Trade (TAST). In more detail, we use the ratio of disembarked slaves in any given state to the total number of slaves disembarked in the US during the TAST as an instrument for the state’s slave share in 1860. The relevance of the instrument is assured by the high correlation between the two variables (almost 0.49). Its excludability can be defended on the ground that it is reasonable to expect that the impact of the share of disembarked slaves on contemporaneous racial educational inequality only runs through the effect on the slave share in 1860 and its subsequent influence on disenfranchisement and education policy. The resulting 2SLS estimates – where slavery is instrumented with the share of disembarked share – confirm the absence of a direct effect of slavery in 1860 on the racial educational gap and that slavery only affects the latter through the initial gap.

As a second step, we proceed by estimating the effect of the initial gap in 1940 on the subsequent evolution of educational inequality using a 2SLS estimator where past slavery in 1860 is employed as an instrument for the initial gap in 1940. While the relevance of the instrument is assured by its previously-reported high correlation with the initial gap, our results from the first step now also support its excludability, since in previous regressions we show that past slavery affects the educational gap only through the initial gap. The 2SLS estimates confirm the impact of the initial gap and thus the persistence of racial educational inequality. These results, taken together with the link we establish between initial gap and slavery, validate the conjecture that the origins of education inequality are indeed very deeply rooted throughout American history.

Once having established the nature of the link between past slavery and racial educational inequality, we turn to its potential association with income growth throughout the 1940–2000 period. Again we proceed in two steps. First, we show that slavery in 1860, again instrumented with the share of disembarked slaves resulting from the TAST, has no direct effect on growth. Second, we can use slavery in 1860 as an instrument for the initial educational gap and show that the latter exerts a negative influence on growth. Thus, we find that slavery matters for growth but once again only indirectly and we identify this channel as racial educational inequality.

To sum up, our results show a persistent effect of the 1940 level of racial educational inequality both on subsequent racial educational inequality and income growth. We also find that past slavery still matters for both outcomes, not directly but indirectly through its influence on the initial gap in attainment between blacks and whites. In other words, those US states that relied more intensively on the use of slave labor still exhibit today larger racial disparities in education and lower growth, through a channel which is represented by the initial gap in attainment, i.e., through the transmission of human capital.

As a by-product of our investigation, we also suggest that past slavery can be used as an instrument for the degree of educational inequality across US states still prevailing at the eve of World War Two, because it affects contemporaneous outcomes only through this indirect channel. Since previous work on the link between inequality and development has been plagued by reverse causality and omitted variables bias, this result is of independent interest and could be exploited in future research.

The paper is organized as follows: Section 2 summarizes the related literature. Section 3 presents the stylized facts and begins our empirical investigation on the determinants of the racial educational gap over the 1940–2000 period through OLS estimates. In Section 4 we illustrate our two-step instrumentation strategy to obtain consistent 2SLS estimates. Section 5 extends the analysis to the determinants of income growth. Section 6 derives our conclusion.

Section snippets

Related literature

This work is related to an emerging literature which is focused on the long-term effect of slavery on the US economy and society. Building on Engerman and Sokoloff, 2005, Nunn, 2008a finds that past slave use is negatively correlated with current economic development for a sample of US counties.4 Over the same sample, Bertocchi and Dimico (2011a) instead show that the main effect of past slavery is on current income

The determinants of the racial educational inequality

We start looking at the dynamics of racial inequality along the educational dimension using a panel of educational attainment across races for the US states in the 1940–2000 period. Table 1 shows the shares of whites and blacks with at least either a high-school education or a bachelor degree. Over the 1940–2000 period whites are on average more educated than blacks. The share of white population with at least a high-school level of education is above 60% against a 47% of black population. The

Empirical strategy

Our results from the previous section show that, over the period under consideration, after controlling for a number of key factors, subsequent levels of the racial educational gap are largely explained by the initial gap and that they are only indirectly affected by slavery through its effect on the initial gap. This suggests that slavery may be excludable in a regression for the educational gap on its initial level. If this were the case, slavery would qualify as a reasonable instrument for

Slavery, the educational gap, and economic growth

Having established a robust link between slavery and education, we now turn to a broader question. Has slavery slowed down the development process in states which made a larger use of this form of labor force? And, if so, through which channels?

In order to shed light on this issue we first look at a direct effect of slavery on economic growth and then we move forward to evaluate its potential indirect effect, which works through educational inequality between races. More precisely, we assess

Conclusion

The legacy of slavery still looms over American society, but debate arose over whether this legacy can still exert a measurable influence on the economic and social achievements of blacks. This paper shows that the contemporaneous degree of racial inequality in education is indeed affected by slavery through its effect on the level of the gap at the eve of World War Two. Over the same period, we also find that income growth is negatively affected by the initial educational disparities between

Acknowledgments

We would like to thank the editors Daniel Berkowitz and Marla Ripoll and an anonymous referee for helpful comments and suggestions. Generous financial support from the Fondazione Cassa Risparmio di Modena and the Italian University Ministry is gratefully acknowledged.

References (49)

  • S.V. Cameron et al.

    The dynamics of educational attainment for blacks, hispanics, and whites

    Journal of Political Economy

    (2001)
  • D. Card et al.

    School quality and black-white relative earnings: a direct assessment

    Quarterly Journal of Economics

    (1992)
  • Eltis, D., 2008. A brief overview of the trans-atlantic slave trade. Voyages: The Trans-Atlantic Slave Trade Database...
  • S.L. Engerman et al.

    Institutional and non-institutionals explanations of economic differences

  • S.L. Engerman et al.

    History lessons: institutions, factor endowments, and paths of development in the new world

    Journal of Economic Perspectives

    (2000)
  • P.V. Fishback et al.

    Narrowing the black–white gap in child literacy in 1910: the roles of school inputs and family inputs

    Review of Economics and Statistics

    (1991)
  • R.W. Fogel et al.

    Time on the cross: evidence and methods

    (1974)
  • M. Frank

    Inequality and growth in the United States: evidence from a new state-level panel of income inequality measures

    Journal of Economic Inquiry

    (2009)
  • O. Galor et al.

    Inequality in landownership, the emergence of human-capital promoting institutions, and the great divergence

    Review of Economic Studies

    (2009)
  • C. Goldin

    America’s graduation from high school: the evolution and spread of secondary schooling in the 20th century

    Journal of Economic History

    (1998)
  • C. Goldin et al.

    The Legacy of US educational leadership: notes on distribution and economic growth in the 20th century

    American Economic Review

    (2001)
  • C. Goldin et al.

    The Shaping of higher education: the formative years in the Unites States, 1890–1940

    Journal of Economic Perspectives

    (1999)
  • C. Goldin et al.

    The great compression: the US wage structure at mid-century

    Quarterly Journal of Economics

    (1992)
  • J.J. Heckman

    The central role of the south in accounting for economic progress of black Americans

    American Economic Review

    (1990)
  • Cited by (42)

    • Population change and the legacy of slavery

      2020, Social Science Research
      Citation Excerpt :

      However, others have taken a structural, place-based approach to understand the lingering societal consequences of this historical institution. This place-based framework suggests that slavery affected places and not just (black) people (Acharya et al., 2016; Bertocchi and Dimico, 2012, 2014; Curtis and O'Connell, 2017; Gouda and Rigterink, 2017; Levernier and White, 1998; O'Connell, 2012; Reece, 2019; Reece and O'Connell, 2016; Vandiver et al., 2006). Not surprisingly, it is within social structures and other dimensions of place, as opposed to individuals, that the legacy of slavery is most persistent.

    • Inequality, crime, and the long run legacy of slavery

      2019, Journal of Economic Behavior and Organization
      Citation Excerpt :

      Recent economic literature has widely investigated the long-term effect of slavery on various economic outcomes. Empirical studies provide evidence on the long-term legacy of slavery income inequality (Bertocchi and Dimico, 2014), economic development (Acemoglu et al., 2002; Nunn, 2008; Maloney and Valencia Caicedo, 2016), racial educational inequality (Bertocchi and Dimico, 2012), and political attitudes (Acharya et al., 2016). More closely related to our work, Gouda and Rigterink (2016) find that the US counties that historically had a higher fraction of slaves are significantly more violent today.

    View all citing articles on Scopus
    View full text