Historical Perspectives on Racial Economic Differences: A Summary of Recent Research

Few public policy debates in the United States are as contentious or as long lasting as those arising from racial economic differences. Historical perspective is essential to these debates because history casts a long shadow - what happened in the past, even the distant past, can affect economic behavior today - and because race is central to so much of the political, social, and economic history of the United States. Race, as the Nobel Prize-winning economist Gunnar Myrdal put it, is the "American dilemma."

Much of the research that I have conducted while associated with the NBER has focused on racial economic differences. For example, my book Race and Schooling in the South, 1880-1950: An Economic History is an extended analysis of the economics of segregated schools in the South prior to the Supreme Court's famous decision in Brown v. Board of Education whose 50-year anniversary is celebrated this year.⁽²⁾ In this summary I briefly discuss my recent work on racial differences, most of which has been conducted jointly with NBER Research Associate William J. Collins, my colleague at Vanderbilt University.

Racial Differences in Schooling

In the United States today black children lag behind their white counterparts in most dimensions of schooling. These gaps have been attributed variously to racial differences in the quality of schooling, family background, neighborhood and other environmental factors, and to cultural biases in testing procedures. Economically, the schooling gaps matter because the American labor market rewards schooling, and these rewards have grown larger over time.

Collins and I⁽³⁾ attempt to provide some historical perspective on contemporary racial differences in schooling. Our work draws heavily on recently available public use samples of various federal censuses, as well as on other public documents. We interpret the evidence in an "analytic narrative" that is based conceptually on a simple model of optimal investment in schooling. The narrative has three principal themes. First, in all the dimensions that the data address, the long-term pattern is one of substantial racial convergence. Second, convergence is not a recent phenomenon; it began long before the Civil Rights Revolution of the 1960s. Third, the South is central to the narrative because, historically, most blacks lived in the South and the educational conditions in the South lagged substantially behind other regions for both races.

Our paper considers a variety of schooling indicators in depth, but the data on illiteracy and school attendance serve to illustrate the major themes. In 1870, the first year for which national data by race were reported, the aggregate racial gap in literacy rates was an astounding 68 percentage points. The gap was so high because the vast majority of blacks at the time were former slaves or their offspring, and literacy was extremely difficult to acquire under slavery. Of the many "treatment effects" of the Civil War, the establishment of schools for black children in the South in the aftermath of the War was perhaps one of the most important, for it enabled successive generations of black children to become literate. Although literacy per se did not require much exposure to formal schooling, the returns to literacy, measured in terms of occupational status (a proxy for income) were quite substantial for blacks - even in the South, where racial oppression and segregation were the norm. There is also some evidence of "pure catch-up," a willingness on the part of black parents to have their children invest in schooling beyond what would have been predicted given the historical circumstances. To be sure, the convergence was not always continuous, especially around the turn of the century when most adult blacks in the South were disenfranchised at the local and state level. However, private philanthropy took up some of the slack as did (later in the century) court action, social activism, and finally, government intervention.

Racial Differences in Housing

Although economic historians and labor economists have long been interested in the historical evolution of racial differences in income and education, less attention has been paid to other types of racial differences in economic status, including housing. We study housing because, in the United States, racial gaps in wealth are much larger than racial differences in income. Although racial gaps exist across all types of assets, those related to housing are particularly salient, because housing equity is a major component of household wealth and African-Americans hold a relatively higher proportion of wealth in owner-occupied housing. Housing is also a major component of private consumption, and housing values reflect both the housing services embodied in the housing unit and access to transportation, employment, retail establishments, security, and various public goods.

Collins and I have written several papers about the long-run evolution of racial differences in housing, all of which drew in one way or another on the public use samples of the U.S. census. In one paper,⁽⁴⁾ we studied secular trends in racial differences in home ownership. African-Americans emerged from slavery with little or no physical wealth but, by 1900, nearly 22 percent of African-American male household heads owned their homes. Considering the initial condition -- near zero wealth in 1870 -- this is an impressive accomplishment. But the rate of black home ownership fell far below that of white household heads at the time -- 46 percent -- implying a racial gap of 24 percentage points. Still, if we control for various correlates of home ownership, such as the age of the household head, literacy and occupational status, and location, then the "unexplained" portion of the racial gap declines to 15 percentage points.

Over the next 40 years there was little overall change in either the black or white homeownership rate and, consequently, in the racial gap. For blacks, homeownership rates did rise during the first decade of the twentieth century, but they fell between 1910 and 1920. The relevant correlate here was the "Great Migration" from the rural South to the urban North; blacks (and whites) living in central cities were far less likely to be homeowners than those living elsewhere. Black homeownership continued to slide between 1920 and 1940, largely because of declines during the Great Depression of the 1930s.

In 1940, the eve of World War II, slightly more than 20 percent of black male household heads were homeowners, compared with 42 percent of white male household heads. The ensuing two decades would witness a vast transformation in American housing, one in which homeownership rates rose substantially for both races. But the gains were larger in absolute terms for whites than for blacks. In 1960, the black homeownership rate stood at 39 percent, while that for whites was 66 percent, implying a larger racial gap. However, if we control for the correlates of homeownership, then the unexplained gap is about the same as in 1940 (or in 1900). Again, the culprit was migration north: migrants were less likely to be homeowners, particularly those migrating to central cities.

In the period since 1960, the racial gap in homeownership among male household heads has narrowed. In 1990, the last year examined in this paper, the racial gap was 19.5 percentage points, compared with 27 points in 1960. Because white homeownership rates were rising over this period, all of the narrowing of the gap reflects a faster pace of growth among black household heads. Moreover, when we control for the correlates of homeownership, the unexplained racial gap fell sharply from 1960 to 1990.

In a second paper,⁽⁵⁾ Collins and I supplement our long-run analysis of home ownership with information on the value of owner-occupied housing. In 1940, the first year for which sample information is available, the black-to-white ratio of the value of owner-occupied housing was 0.37. The ratio then increased sharply over the next three decades, to 0.62 in 1970, reflecting a narrowing of the racial gap in housing characteristics that affect value, such as the number of rooms or the presence of indoor plumbing. But from 1970 to 1990 the aggregate national ratio was essentially unchanged, while that for central cities, where most black households resided, declined sharply.

In further analysis of the deterioration in the relative value of black-owned housing in the 1970s, we examine the correlation between the black-white housing value ratio and the level of residential segregation. Prior to 1970, the black-white ratio was either higher in heavily segregated metropolitan areas or essentially unrelated to the level of segregation. In the 1970s, the correlation become strongly negative; in other words, the deterioration in the relative value of black-owned housing was most severe in cities that were highly segregated.

A variety of economic models suggest that high levels of racial segregation can lead to a "downward spiral" in economic outcomes for blacks in response to a negative shock. The best known among them is that of David Cutler and Edward Glaeser.⁽⁶⁾ Using 1990 census data, they show that increases in residential segregation lead to worse economic outcomes for blacks, a phenomenon known as "bad ghettos." But their analysis leaves open the question of whether ghettos were always bad. Using similar empirical methods, Collins and I demonstrate that, although bad ghettos certainly existed prior to 1970, the process intensified during the 1970s and 1980s.⁽⁷⁾

Our finding that the black-to-white ratio of property values in central cities fell in the 1970s is consistent in timing with the emergence of "bad ghettos" but raises the obvious question: what caused this emergence? We are not the first to consider this question, and it is unlikely that a single "smoking gun" is responsible, or that all of the causes can be separately identified and measured. But perhaps some can be. In our work, Collins and I have explored the effects of one possible trigger: race-related civil disturbances or "riots."

Although the United States has experienced many race riots throughout its history, those occurring in the 1960s were unprecedented in frequency and scope. Social scientists, though long interested in the causes of the 1960s riots, have done relatively little work of an econometric nature on their consequences. In two recent papers, Collins and I use census data to examine the impact of the riots on labor and housing market outcomes for blacks in a standard "difference-in-difference" econometric framework; that is, we compare changes between 1960 to 1970, and 1960 to 1980, in an outcome variable (for example, median black family income) in cities that experienced a severe riot versus cities that did not.⁽⁸⁾ In terms of injuries, deaths, or destruction of property, the severity of riots varied considerably, and it is important to take this into account in the analysis.

One key issue is whether the occurrence of a riot in a particular city might be endogenous to the outcome under study. For example, if riots were more frequent in cities in which black economic prospects in 1960 were especially poor, then the difference-in-difference estimator might produce a biased estimate of the treatment effect. However, the bulk of the work on the causes of the riots suggests that few if any reliable predictors of riot activity can be measured at the city level, other than region (the South had fewer riots) and the absolute size of the black population, both of which we control for. We also consider two-stage least squares estimates in which local government organization (the use of a city manager) and rainfall in the period around the time of the assassination of Martin Luther King (rainfall substantially reduces the likelihood of a riot) serve as instrumental variables. Our empirical work relies on city-level data from the 1950-80 population censuses and individual-level data from the 1970-80 census samples.

We find that the occurrence of a severe riot had economically significant negative effects on blacks' income and employment prospects, and that these effects appear to have been larger in the long run (1960-80) than in the short run (1960-70). For example, the negative effect on median black family income was on the order of 9 percent in the 1960s. The value of black-owned property was also adversely affected in the 1960s by the occurrence of a riot, with little or no rebound in the 1970s. Individual-level data from the census samples suggest that the racial gap in property values widened in the 1970s in cities that experienced even moderately severe riot activity.

The exact conduit though which these negative effects emerged is next to impossible to identify with the data at hand, but it is straightforward to speculate about the likely channels. Property (and personal) risk was heightened by riots; qualitative evidence suggests that insurance premiums increased after a riot. Taxes for police and fire protection may have increased, and some riot cities had difficulty placing municipal bonds. Retail establishments that were burned or damaged might not reopen, businesses and households might move away, and so on. Some of the negative effects could have been (and were) offset by outside assistance, but evidently on balance the negative effects predominated. Moreover, because the occurrence of a major riot was national news, it is likely that our empirical strategy underestimates the negative effects. In future work, we plan to examine the effects at the census-tract level, and also explore other possible impacts, notably those on crime and local politics. Crime rates are known to have increased in the 1970s but the relationship between the occurrence of a riot and subsequent crime remains to be explored. A number of American cities elected African-American mayors for the first time in their histories in the 1970s but whether the riots speeded up or hindered the likelihood of electing a black mayor is unclear.

The Civil War and Black Economic Progress

Prior to the Civil War the vast majority of African-Americans were enslaved. With the end of the Civil War came the end of the slavery, and with it, the first prospects for economic advancement among former slaves. But the pace of black economic advance was hindered by the fact that, in the aftermath of the war, most blacks lived in the South, and the South was undeniably poor. At the turn of the twentieth century, for example, per capita income in the South was approximately half the national average.

Economic historians have wondered about the causes of southern poverty, especially the role played by the Civil War. In the two decades prior to the War per capita incomes in the South grew at about the national average. But in the aftermath of the War, southern per capita incomes fell sharply, both absolutely and relative to the national average, and recovery was slow. A variety of explanations have been proposed to account for the decline and the slow pace of recovery, but there is a lack of consensus on the relative importance of these different explanations.

My approach to this debate has been to disaggregate the effects of the War by focusing on the components of per capita income -- namely, factor prices and per capita factor supplies. Focusing on the components, particularly on factor prices, is useful, because additional data can be brought to bear and, more importantly, because different explanations often imply very different changes in factor prices.

In a recent paper,⁽⁹⁾ I examine the impact of the Civil War on wages in the South relative to the North. Many blacks entered the wage labor market after the War, either on a part-time or full-time basis, so data on wage movements are particularly relevant. Compared with pre-war levels, nominal wages in the South fell sharply relative to the North in the immediate aftermath of the War. And, such declines occurred for a broad range of occupations. While there was some recovery in the 1880s, agricultural distress in the 1890s led to further erosion in Southern relative wages. I also show that real wages in the South fell, but that the declines were smaller in magnitude, because the cost of living fell as well. One of the more prominent explanations of the post-bellum decline in Southern per capita income is an exogenous reduction in per capita labor supply in the South. However, my results suggest that this cannot be the dominant explanation because, if it were, relative wages in the South would have risen, not fallen.

In ongoing work with my Vanderbilt colleague William Hutchinson,⁽¹⁰⁾ I examine changes in wage-rental ratios in the South relative to the North after the War. Although wages fell in the South, interest rates rose, resulting in sharp declines in the cost of labor compared with the cost of capital. Simple economic theory predicts that capital intensity should have decreased in the South in response to this change in relative factor prices. Using establishment level data from the 1850-80 censuses, Hutchinson and I demonstrate that manufacturing establishments in the South did experience a sharp decline in capital intensity after the War relative to establishments outside the region. Our preliminary results also suggest that manufacturing labor productivity fell in the South relative to the North after the War, and that the decrease in relative labor productivity can be accounted for fully by the reduction in relative capital intensity.

2. R. A. Margo, Race and Schooling in the South, 1850-1950: An Economic History, Chicago: University of Chicago Press, 1990, NBER Long-Term Factors in Economic Development monograph series.

3. W. J. Collins and R. A. Margo, "Historical Perspectives on Racial Differences in Schooling in the United States," NBER Working Paper No. 9770, June 2003, forthcoming in E. Hanushek and F. Welch, eds., Handbook of the Economics of Education, New York: Elsevier.

4. W. J. Collins and R. A. Margo, "Race and Home Ownership: A Century-Long View," Explorations in Economic History, 37 (2001), pp. 68-92 (revised version of NBER Working Paper No. 7277, August 1999).

5. W. J. Collins and R. A. Margo, "Race and the Value of Owner-Occupied Housing, 1940-1990," Regional Science and Urban Economics, 33 (2003), pp. 255-86 (revised version of NBER Working Paper No. 7749, June 2000).

6. D. M. Cutler and E. Glaeser, "Are Ghettos Good or Bad?" Quarterly Journal of Economics, 112 (1997), pp. 827-72.

7. W. J. Collins and R. A. Margo, "Residential Segregation and Socioeconomic Outcomes: When did Ghettos Go Bad?" Economic Letters, 69 (2000), pp. 239-43.

8. W. J. Collins and R. A. Margo, "The Labor Market Effects of the 1960s Riots," in W. Gale and J. Pack, eds. Brookings-Wharton Papers on Urban Affairs 2004, pp. 1-24. Washington, DC: The Brookings Institution (revised version of NBER Working Paper No. 10243, January 2004); and "The Economic Aftermath of the 1960s Riots: Evidence from Property Values," NBER Working Paper No. 10493, May 2004.

9. R. A. Margo, "The North-South Wage Gap, Before and After the Civil War," in D. Eltis, F. Lewis, and K. Sokoloff, eds. Slavery in the Development of the Americas, pp. 324-51, New York: Cambridge University Press, 2004 (revised version of NBER Working Paper No. 8778, February 2002).

10. W. Hutchinson and R. A. Margo, "The Impact of the Civil War on Capital Intensity and Labor Productivity in Southern Manufacturing," in progress.