%20-%20https://sketch.com%20--%3e%3ctitle%3e3ECBBF50-FD08-412A-943D-2CFC5E025A16%3c/title%3e%3cdesc%3eCreated%20with%20sketchtool.%3c/desc%3e%3cg%20id='Stylesheet'%20stroke='none'%20stroke-width='1'%20fill='none'%20fill-rule='evenodd'%3e%3cg%20transform='translate(-926.000000,%20-668.000000)'%20id='BUTTONS'%3e%3cg%20transform='translate(100.000000,%20597.000000)'%3e%3cg%20id='BTN-/-RIGHTS'%20transform='translate(816.000000,%2065.000000)'%3e%3cg%20id='ICON-/-Rights'%20transform='translate(10.000000,%206.000000)'%3e%3cpath%20d='M13.5752841,0%20L1.34943182,0%20C0.59375,0%200,0.586666667%200,1.33333333%20L0,22.6666667%20C0,23.4133333%200.620738636,24%201.34943182,24%20L17.6505682,24%20C18.40625,24%2019,23.3866667%2019,22.6666667%20L19,5.36%20L13.5752841,0%20Z%20M13.5752841,1.89333333%20L17.0838068,5.36%20L13.5752841,5.36%20L13.5752841,1.89333333%20Z%20M1.34943182,22.6666667%20L1.34943182,1.33333333%20L12.1988636,1.33333333%20L12.1988636,5.36%20C12.1988636,6.10666667%2012.7926136,6.69333333%2013.5482955,6.69333333%20L17.6235795,6.69333333%20L17.6505682,22.6666667%20L1.34943182,22.6666667%20Z'%20id='Shape'%20fill='%23595959'%20fill-rule='nonzero'%3e%3c/path%3e%3ccircle%20id='Oval'%20stroke='%23595959'%20stroke-width='1.5'%20fill='%23595959'%20cx='9.5'%20cy='13.5355339'%20r='5.5'%3e%3c/circle%3e%3cpolyline%20id='Path-2'%20stroke='%23FFFFFF'%20stroke-width='1.5'%20transform='translate(9.500000,%2013.035534)%20rotate(45.000000)%20translate(-9.500000,%20-13.035534)%20'%20points='8%2016.0355339%2011%2016.0355339%2011%2010.0355339'%3e%3c/polyline%3e%3c/g%3e%3c/g%3e%3c/g%3e%3c/g%3e%3c/g%3e%3c/svg%3e)
America’s renewed reckoning over racial inequality has brought fresh attention to the economic history literature on persistent gaps in wealth and income.Footnote 1 In accounts concerning the origins and evolution of racial disparities, scholars have long seen the post-Civil War era of Reconstruction (1865–77) as a critical juncture (Du Bois 1935; Myrdal Reference Myrdal1944; Higgs Reference Higgs1977; Ransom and Sutch 1977; Jaynes Reference Jaynes1986; Foner 1988). In particular, the nation’s failure to provide the formerly enslaved population with the resources and protections required to participate fully in economic and political life hindered Black Americans’ income, wealth, and human capital accumulation and perpetuated disparities in economic well-being that are still present today.
Explanations of the ways in which Reconstruction-era policy choices undermined the economic status of Black Americans and their descendants often focus on the distribution of land ownership.Footnote 2 The Civil War and Reconstruction ended without implementing a plan to distribute land or compensation to those formerly enslaved. Consequently, even by 1880, relatively few Black households owned real property. Collins and Margo (2011, appendix table 2) estimate that approximately 14 percent of Black households owned homes in 1880, compared to 55 percent of White households. At this time, 85 percent of Black Americans lived in rural areas. Land ownership offered an escape from farm tenancy, sharecropping, and wage labor, affording Black households more control of their daily lives and the fruits of their production. Accordingly, they made land their paramount economic concern. Eric Foner explains that “The desire to escape from white supervision and establish a modicum of economic independence profoundly shaped blacks’ economic choices during Reconstruction… Above all, it inspired the quest for land of their own” (1988, p. 104, emphasis added).
This paper provides new evidence on the distributions of Black and White landholdings at the end of Reconstruction and on the subsequent intergenerational transmission of economic status. We build a novel dataset of fathers and sons with national coverage, including thousands of hand-linked and transcribed records that bridge the Census of Agriculture and Census of Population manuscripts in 1880. Such record linkage is essential for characterizing the relationship between landholding in one generation and economic status in the next because the 1880 Census of Population contains information on occupations but no information on property ownership, let alone farm characteristics, whereas the Census of Agriculture manuscripts, which provide the most consistent micro-level and national-in-scope data on farms, contain no information on the farmer’s race or children.Footnote 3 The detailed view of differences in status within the occupation of “farmer” is a key contribution of this dataset. To capture the intergenerational transmission of economic status, we link children from all households in 1880, including those of non-farmers, to their records in the 1900 Census of Population manuscripts, where we can observe their occupation, literacy, homeownership status, and more.
We first show that the farm ownership rate and the distributions of farm acreage and value differed starkly for Black and White households in 1880. It is immediately clear that allocating 40 acres of farmland to Black households—as proposed during and after the Civil War but not implemented (Cox Reference Cox1958; Oubre 1978; Foner 1988)—would have greatly increased Black Americans’ landholdings and wealth relative to what they had actually attained by 1880. Although some Black households owned farmland in 1880 (approximately 12 percent in our sample), most men worked as sharecroppers or wage laborers.
We then investigate the connection between fathers’ land ownership and farm size in 1880 and sons’ outcomes in 1900, showing that the returns in terms of sons’ homeownership and literacy were substantial. For instance, conditional on age, sons of landowning Black farmers achieved a homeownership rate that was 7 percentage points higher than for sons of Black laborers, whereas sons of farmers who did not own land saw no such advantage. Sons of landowning farmers were also 11 percentage points more likely to be literate in 1900 than the sons of laborers, and 13 percentage points more likely to be literate than the sons of farmers who rented. We also show that sons of farmers in the upper quintiles of owned acreage or farm value fared better in terms of literacy and homeownership than sons of landless farmers and, often, better than those in the lower quintiles of owned acreage or value. These are fundamental insights about the first generations of Black households in the post-Civil War period, which are available only in datasets that link information drawn from both the agricultural and population censuses.
It is important to be clear that our measures of sons’ outcomes conditional on fathers’ landholdings or occupations cannot be given a simple causal interpretation, nor is that what we aim to provide.Footnote 4 The connections between the economic status of one generation and that of the next are manifold, and the intergenerational mobility literature recognizes that statistical associations may reflect investments in human capital, social connections, local resources and environments, and more (inter alia, Loury Reference Loury1977; Becker and Tomes Reference Becker and Tomes1986; Solon Reference Solon, Ashenfelter and Card1999; Margo 2016). In the context we study, all of these investments were affected in complex ways by the history of slavery and ongoing discrimination and violence against Black Americans. Although we do not identify specific causal mechanisms, our contribution is to show whether and how strongly differences in economic status among Black households, especially with respect to farm ownership and fathers’ occupations, were transmitted to the next generation in the wake of Reconstruction.
We find that Black families with more economic resources transmitted substantial advantages to their children in the form of both property ownership and human capital accumulation. Moreover, we show that local fixed effects do not greatly weaken these relationships. Therefore, we conclude that the advantages were likely due primarily to “within family” mechanisms rather than differences across localities in Black Americans’ opportunities.
Despite these advantages, the limits of landownership as a springboard for the next generation’s prosperity are also clearly evident. Even for Black sons of landowning farmers, average rates of literacy and homeownership in 1900 were far lower than for White sons. In fact, sons from the top quintiles of Black farm-owning families (by value or acres) had much lower literacy rates and slightly lower homeownership rates than sons of White farmers who did not own farmland. Moreover, in terms of the sons’ occupational status in 1900, we find that the advantages associated with Black fathers’ land ownership were small compared to the magnitude of the Black-White gap. We interpret this as a reflection of the era’s severe limits on Black workers’ educational and labor market opportunities.
Our work contributes to several strands of the literature on American economic development, especially regarding the economic status of Black Americans since emancipation. First, our work complements that of Miller (Reference Miller2020), which compares freedmen’s outcomes in the Cherokee Nation (in Northeast Oklahoma), where they were granted land after the war, to outcomes elsewhere in the South. Miller (Reference Miller2020) finds that Black farm acreage, ownership, livestock value, and crop production were higher in the Cherokee Nation than elsewhere in the South, and some of these advantages were likely transmitted to the next generation (e.g., higher school attendance rates). While similar in motivation, this paper provides a different and geographically broader empirical perspective on the experiences of freedmen and their children in comparison with the White population.Footnote 5
Second, the paper complements a long-standing literature on wealth inequality in the post-Reconstruction period. Much of this research is based on geographically aggregated tax data and, therefore, does not observe intra-household transmission dynamics (Du Bois Reference Du Bois1901; Higgs 1982; Margo Reference Margo1984; Spriggs Reference Spriggs1984; Derenoncourt et al. Reference Derenoncourt, Hyun Kim, Kuhn and Schularick2022). The paper also contributes to the closely related literature on the long-run persistence of racial inequality (Myrdal Reference Myrdal1944; Higgs Reference Higgs1977; Ransom and Sutch 1977; DeCanio Reference DeCanio1979; Oliver and Shapiro Reference Oliver and Shapiro1995; Margo 2016). Farming and land ownership (or lack thereof) are prominent themes in this research, especially for the nineteenth and early twentieth centuries. In this regard, our data collection efforts—specifically the links made between the Census of Agriculture and the Census of Population manuscripts—are akin to those undertaken by Parker and Gallman (1992), Bateman and Foust (1992), and especially Sutch and Ransom (2007).Footnote 6 Our sample construction differs from theirs in important ways, most notably in the provision of forward linkages to 1900 for intergenerational outcomes. But the extensive research based on these datasets, like our paper, highlights the value of the agricultural census for understanding American economic development and inequality.
Finally, the paper contributes to the broad literature on intergenerational mobility in American history (inter alia, Aaronson and Mazumder Reference Aaronson and Mazumder2008; Long and Ferrie Reference Long and Ferrie2013; Olivetti and Paserman 2015; Bleakley and Ferrie Reference Bleakley and Ferrie2016; Ager, Boustan, and Eriksson 2019; Tan Reference Tan2023), including work that specifically addresses racial differences (Hertz Reference Hertz, Bowles, Gintis and Osborne2005; Mazumder Reference Mazumder2014; Collins and Wanamaker Reference Collins and Wanamaker2022; Derenoncourt Reference Derenoncourt2022; Jácome, Kuziemko, and Naidu Reference Jácome, Kuziemko and Naidu2022; Ward Reference Ward2023). It deepens these literatures by providing new insights into the economic outcomes of formerly enslaved men and women and their children, with unprecedented detail on variation in economic status within the large population of farmer-headed households. It also expands beyond the literature’s traditional emphasis on occupational status transmission to bring other important variables into view, such as literacy and property ownership.
CONTEXT, DATA, AND ANALYTICAL FRAMEWORK
Historical Context
After the Civil War, formerly enslaved men and women sought land ownership as a bulwark to protect their autonomy, whereas southern planters sought to re-establish their control over Black labor and resume large-scale production of staple crops. Foner points out that southern Whites were influenced by the earlier experiences of Haiti and the British Caribbean: “Planters quickly concluded that their ability to control black labor rested upon maintaining their own privileged access to the productive land of the plantation belt” (1988, pp. 134). Approximately 90 percent of the Black population resided in the South, where the economy was predominantly rural and agricultural: 77 percent of southern men worked in agriculture in 1870, and 68 percent did so 30 years later.Footnote 7 The importance of land in such a setting—acquiring it, holding it, and expanding production upon it—was clear. Reviewing the first decades after Emancipation, Du Bois emphasized, “Perhaps there could be found no other single index of the results of the struggle of the freedman upward so significant as the ownership of land” (1901, p. 648).
In March 1865, Congress created the Freedmen’s Bureau and authorized it to divide abandoned and confiscated Confederate lands into 40-acre plots for rent and eventual sale to those formerly enslaved. No sooner had the Freedmen’s Bureau begun its work than President Andrew Johnson started to unwind its efforts. Johnson issued thousands of presidential pardons to wealthy Confederate supporters, thus restoring their sizable landholdings and dispossessing Black farmers. Foner concludes, “Johnson had in effect abrogated the Confiscation Act and unilaterally amended the law creating the Bureau. The idea of a Freedmen’s Bureau actively promoting Black landownership had come to an abrupt end” (1988, p. 161).
The following year, Congress passed the Southern Homestead Act, which opened public lands in the South to settlement. The formerly enslaved population and other residents who had remained loyal to the Union were given preferential access until 1867. This initiative did not greatly increase Black landownership for multiple reasons: the land available from public holdings was of low quality for farming; most Black men had already agreed to labor contracts that required them to work until the end of 1866; land offices were poorly staffed and operated; and most Black households had little or no capital to support themselves while starting a farm (Oubre 1978, pp. 183–88). Once they were recognized as citizens under the Fourteenth Amendment, homesteading elsewhere in the United States under the 1862 Homestead Act was possible, but relatively few Black households pursued opportunities far from the South (Edwards, Friefeld, and Eckstrom 2019).Footnote 8
Without land of their own, the overwhelming majority of Black Americans began the post-emancipation period with no resources other than their ability to labor.Footnote 9 White households, on the other hand, continued to control the most productive agricultural land, and those in the South continued to benefit from employing a large pool of landless Black workers. In the 1870 Census of Population, the last census to record real estate asset values, approximately 71 percent of all southern Black men between the ages of 18 and 60 were laborers (agricultural or otherwise). Although 17 percent were recorded as “farmers,” approximately 90 percent of these farmers apparently did not own land (real estate assets are recorded as zero), and most likely worked as sharecroppers.
From this starting point, the formerly enslaved population worked toward acquiring land, typically by ascending the “agricultural ladder” from laborer to cropper, tenant, and finally owner. To advance these efforts, Logan (Reference Logan2020) documents a push by Black politicians elected during Reconstruction to raise property taxes, a policy favored by Black constituents to pressure White landowners to sell off unimproved land. The Black population’s gains in property ownership between 1870 and 1900, as documented by Du Bois (Reference Du Bois1901), Higgs (1982), and Margo (Reference Margo1984), are remarkable when viewed against the backdrop of discrimination and violence that characterized the era. At the national level, Collins and Margo (2011) show that Black homeownership and farm ownership rates increased from 1870 to 1900 even as the White homeownership rate declined due to ongoing urbanization.
Opportunities for economic advancement outside the agricultural sector were limited for Black workers in the decades after Reconstruction. A nascent southern manufacturing sector often excluded Black workers (Greene and Woodson Reference Greene and Woodson1930; Myrdal Reference Myrdal1944; Wright Reference Wright1986), as did northern manufacturers, as long as White workers, including immigrants, were readily available. Possibilities for advancing upward in any industry, for instance into supervisory positions, were also limited for Black workers (Johnson Reference Johnson1943; Sundstrom Reference Sundstrom1994; Mohammed and Mohnen 2021). Moreover, Black men lost ground in some lines of skilled crafts work in the late nineteenth century (Myrdal Reference Myrdal1944, ch. 13). Although some acquired literacy, few of the formerly enslaved had the opportunity to attain much formal education (Williams Reference Williams2006), and their children attended racially segregated and poorly funded schools (Margo 1990). Professional positions were generally limited to those serving Black clientele (e.g., teachers and clergy).
Post-Civil War disparities in landholding are salient in the overarching economic history of Black-White disparities in income and wealth. Our paper’s goal is to explore these disparities in more depth than previously possible, with an eye toward intergenerational transmission and, thus, the perpetuation of racial inequality. Although the Black population was geographically concentrated in the South during this period, our dataset and presentation of results emphasize a national perspective. This provides a clearer view of overall Black-White disparities in the late nineteenth century, a view that includes the vast majority of the White population, which lived outside the South and accumulated land and wealth there. It is also in keeping with much of the literature on intergenerational mobility, which takes a national perspective. We will also provide some “within-South” perspectives. The South was a relatively poor and isolated region in this period (Wright Reference Wright1986), and the within-South comparisons highlight disparities in landholdings and mobility even among families from the same region.
Dataset Construction
We began with a full count dataset of the 1880 Census of Population (Minnesota Population Center 2017).Footnote 10 We then linked all sons aged 0 to 17 who were living with their father or stepfather in 1880 to their records in the 1900 Census of Population, made available by Ancestry.com through IPUMS (Ruggles et al. Reference Ruggles, Katie Genadek, Grover and Sobek2015). We followed recommendations from Bailey et al. (Reference Bailey, Cole, Henderson and Massey2020) by using the intersection of two samples that we linked using different methods.Footnote 11 Based on their ages and places of birth, the vast majority of the sample’s Black sons were born free to previously enslaved parents—this is essentially the first generation of post-emancipation young Black men. Some of the data described here are derived from Collins and Wanamaker (Reference Collins and Wanamaker2022), but with additional transcription from the agricultural census manuscripts and much more analysis of farm-level data in 1880 to glean insights into this critical period. In 1880, we can observe parents’ occupation, race, age, and location; we cannot observe homeownership or literacy in this sample.Footnote 12 In 1900, we could observe sons’ occupation, location, homeownership, and literacy status; we could not observe the value of homes or farms.Footnote 13
To sharpen our characterization of parents’ socioeconomic status, we linked a random sample of farmer fathers, drawn from the 1880-to-1900 linked sample of Census of Population records described previously, to the 1880 Census of Agriculture records. We oversampled Black farmers. This linkage to the Census of Agriculture allows us to discern the farmer’s tenure status and provides information on farm acreage (improved and unimproved), farm value, livestock value, and other characteristics. We can then distinguish between farm owners, renters for “fixed money” (cash), and renters for “shares of production,” and we can segment farmers by the value and size of their farm ownership. The links to the 1880 agricultural census must be made manually. Undergraduate researchers established the links based on names and detailed locations using digitized and searchable manuscripts available on the Ancestry.com website.Footnote 14 The final working sample includes 4,976 matched White farmer fathers and 3,276 matched Black farmer fathers, plus 299,847 Black and White non-farmer fathers. All analysis samples are weighted to be representative of fathers in 1880.Footnote 15 Fathers who reported being farmers in the 1880 population census but who were not found in the agriculture census are not included in the analyses.Footnote 16
Measures of Sons’ Outcomes in 1900
Homeownership: Homeownership in 1900 is a binary indicator for whether the son is a household head and living in owner-occupied housing. For those who were farmers in 1900, it is likely that a homeowner also owned at least some farmland. For both farmers and non-farmers, homeownership may reflect labor market success, the accumulation of savings, and perhaps inheritance or inter vivos transfers. This variable must be interpreted with some care because homeownership was higher in rural than urban areas, but urban migration sometimes led to higher-paying jobs. Sons who are not household heads in 1900 are omitted from the analysis.Footnote 17
Human Capital: We use literacy as a proxy for human capital accumulation. The census did not record the highest grade of attainment before 1940, leaving us with a basic but still informative measure of educational attainment. Approximately 58 percent of the Black population, ages 20 to 40, could read and write in 1900.
Occupation: In the absence of individual income measures in the 1900 census, we rely on occupations to characterize labor market outcomes. An indicator of laborer status is a measure of whether the son worked on the lowest rung of the occupational distribution. We group farm and non-farm laborers in this category because both groups had low compensation and because it was not uncommon for laborers to move between sectors. For some analyses, we also include indicators for “blue collar” and “white collar” status, where “blue collar” refers to semi-skilled or skilled craft workers.
Income Rank: A supplementary measure of labor market outcomes is the son’s occupational income rank. This is based on a methodology described in full in Collins and Wanamaker (Reference Collins and Wanamaker2022), which takes account of detailed occupations, race, and region.Footnote 18 For farming sons, our income assignments were differentiated based on homeownership status in 1900. Based on these scores, we then rank the sons in the national distribution of scores (i.e., we rank all linked sons against the complete count sample for 1900). Although these ranks go a step beyond “laborer status” in characterizing labor market outcomes, they do require strong assumptions, and a clear shortcoming is that they cannot register individual-level income variation within occupation/race/region cells.
Analytical Framework
We use the new dataset to answer fundamental questions about whether and how Black households’ economic status in 1880 was transmitted to their sons’ outcomes in 1900. We begin with a sample of farmers’ sons. We use detailed information on farm size, farm value, and livestock value to categorize farming fathers into one of six groups. When we use farm acreage, a “zero acreage” farmer is one who rents for cash or rents for shares of production, and the five remaining categories reflect quintiles of the acreage distribution among farm owners (separately by race). When we use farm value to categorize farmers, we assume those who “rent for cash” own the value of their livestock but not the land, and that those who “rent for shares” do not own the farm’s livestock or land.Footnote 19 Our assumption, given our focus on Black men, is that most farmers who did not own their farm and did not rent for cash were sharecroppers, or approximately as well off as sharecroppers. This distinction is coarse, but it reflects a real difference in their average economic status. The farm value variable may capture differences in economic status and resources that are missed by the acreage count. A “zero value” farmer in this cut of the data is one who rents for share or rents for cash on a farm with no livestock; the five remaining categories reflect quintiles of the farm value distribution, including the value of livestock.
We then widen the scope of comparison by including sons of non-farming fathers. In this framework, the fathers’ categories correspond to broad occupational groups. We still differentiate between sons of farm owners and renters, and then the other fathers’ categories include unskilled laborers, white-collar occupations, and blue-collar occupations.
We age-adjust the data to avoid differences in average outcomes that are due solely to differences across groups in the ages at which we observe the household head’s status in 1880. Following Alsan and Wanamaker (Reference Alsan and Wanamaker2018), we separately standardize the sample of farming fathers and all fathers by reweighting the observations within each status category to mimic the age distribution of all fathers in the universe in question.Footnote 20 So, for example, if fathers who rent for shares are younger than fathers in the top quintile of farm acreage, the age adjustment will result in comparisons of sons’ outcomes that are pre-adjusted for these differences. Later, we refer to regression analyses that account for age differences and other potential confounders using fixed effects.
Most of the recent research on race and intergenerational mobility focuses on the twentieth century and attempts to estimate intergenerational income elasticities or father-son “rank-rank” correlations. In work based on historical samples, in which individual-level income is not observable, intergenerational persistence estimates usually entail assigning fathers and sons average or median income scores by occupation (Collins and Wanamaker Reference Collins and Wanamaker2022; Jácome, Kuziemko, and Naidu Reference Jácome, Kuziemko and Naidu2022; Ward Reference Ward2023). These score-based correlations are not directly comparable to those derived from late twentieth-century datasets, in which individual-level income is observable. But they are provided in much the same spirit of quantifying nationwide intergenerational persistence with a single parameter.
This paper is novel in that it is centered on the late nineteenth century, and it aims to provide the first intergenerational perspective that incorporates detailed information about farms. We emphasize a non-parametric view that is centered on categories defined by fathers’ landholdings (for children of farmers) or a combination of categories based on fathers’ farm tenure and a broad set of non-farm occupations (for all children). We do not estimate linear correlations between father-son occupational status, which requires strong assumptions in the assignment of scores and the functional form of status transmission. Therefore, there is not a single elasticity of income persistence that connects this paper’s results to those in the modern literature or in recent occupation score-based analyses.Footnote 21 Instead, we summarize sons’ outcomes within categories of fathers’ status, and we examine a broader set of outcomes, ranging beyond occupational status (e.g., literacy, homeownership, and more).
In sum, our approach complements the existing intergenerational literature by focusing on the decades immediately after Reconstruction, getting inside the heterogeneous category of farmers, and analyzing a range of relevant outcomes beyond occupation scores. Where they overlap, our results are broadly consistent with those from other studies, which we discuss after presenting our main results. However, our aim in this paper is to extend intergenerational analyses into relatively unexplored areas.
Caveats
Our data have limitations. First, our characterization of farm ownership in 1880 is based on information from the Census of Agriculture manuscripts. We do not observe households’ non-agricultural property ownership, though we do observe the sons’ homeownership status in 1900 regardless of their farm or non-farm residence. Given the concentration of Black households in rural areas and the long-standing focus on agricultural land distribution policies, we believe our dataset speaks to a crucially important issue in the history of American racial disparities. Yet we want to highlight that non-agricultural property also merits attention in future research, especially as urbanization increased in the twentieth century.Footnote 22 In addition, it is important to note that if a farm was rented, the census recorded the operator’s name, but not the owner’s name. Given that White households disproportionately owned the stock of rented land, tallies of owned-and-operated farm acreage and value will understate the Black-White gap in total farmland owned.
Second, as in all linked datasets, we must address selection into linkage relative to the base population of fathers with co-resident sons. As mentioned earlier, we accomplish this through reweighting, as recommended in Bailey et al. (Reference Bailey, Cole, Henderson and Massey2020). But differences in unobservables, such as sons’ educational attainment, health, or other aspects of human capital, may also influence linkage probabilities and positively bias the sample. There may also be a positive bias in our ability to link farmers to their farms in the 1880 agricultural census manuscripts.Footnote 23
Third, it is important to keep in mind that the linked sample consists of sons who resided in households headed by their fathers (or stepfathers) in 1880, which makes the characterization of intergenerational mobility more straightforward than otherwise. Although the vast majority of children resided in households headed by their fathers, it is notable that Black children were more likely than White children to reside in households that were not headed by their fathers. Moreover, this selection issue might cut in different ways for different groups. For instance, Black children in non-father-headed households tended to reside with heads that were more literate than those of other Black children, whereas White children did not.Footnote 24 Additional research into these aspects of household formation and their implications for children’s resources and outcomes would be valuable.
Fourth, it is plausible that staying in one place may be correlated with a higher probability of selection into linked samples (e.g., if recent migrants are more likely to be missed by enumerators). We do not have direct evidence on this, but it is useful to consider the issue because a father’s land holdings might lead to stronger local attachments for sons. If so, this may distort the distribution of households across categories of fathers’ status and, thus, overall mobility rates. However, we would expect this to have limited implications for sons’ average outcomes within categories of fathers’ status, which is what our analysis of mobility emphasizes. Alternatively, if men with higher skill levels, conditional on fathers’ status, are more likely to be matched over time, then it is possible that the linked sample’s average outcomes for sons from each category of fathers overstate the true outcomes. If this selection bias is stronger for sons from lower-status households, it could differentially tilt the relationships depicted and lower the estimated returns to fathers’ status.
Finally, despite our effort to limit false matches by adopting a conservative linkage procedure, remaining mismatches might matter. For instance, an 1880 son who is matched to the wrong adult in 1900 might have his occupation, homeownership, and literacy mistaken. This may flatten the observed relationship between father and son’s status. Similarly, Ward (Reference Ward2023) emphasizes that fathers’ occupations are sometimes recorded inconsistently. Our emphasis on fathers’ broad occupational categories might avoid placing too much weight on detailed occupational codes. But there is scope for error, which again may flatten the observed relationship between fathers’ and sons’ outcomes.Footnote 25
RESULTS
Racial Differences in Landholding
We begin with a tabulation of occupations and landholding data, shown in Table 1.Footnote 26 Few Black fathers in our linked sample owned farms in 1880—approximately 12 percent compared to 44 percent of White fathers. This means that only about 25 percent of the sample’s Black farmers owned land, compared to nearly 80 percent of White farmers. This reflects both the legacy of slavery and the failure of plans for land distribution to those formerly enslaved, yet it is also evidence of the determination of many Black households to ascend the “agricultural ladder” and achieve a measure of economic independence. If not farmers, Black fathers were concentrated in the “laborer” category (43 percent), whereas White men were more concentrated in white-collar and skilled or semi-skilled blue-collar jobs.
OCCUPATION AND LAND OWNERSHIP DISTRIBUTION AMONG 1880 FATHERS IN LINKED SAMPLE
Notes: “White collar” includes professional, clerical, and sales occupations. “Blue collar skilled” includes craftsmen primarily and certain service occupations (e.g., policemen and detectives). “Blue collar semi-skilled” includes operatives primarily and certain service occupations (e.g., barbers, bartenders). Laborer includes both blue-collar laborers and farm laborers. The top panel reflects a weighted summary of a sample containing 282,938 White father/son pairs and 25,025 Black father/son pairs. The second panel reflects 4,934 White observations and 3,182 Black observations. The third panel reflects 3,963 White observations and 787 Black observations.
Sources: Farm ownership and rental status are ascertained from the 1880 Census of Agriculture manuscripts. Other occupational categories follow the 1880 Census of Population information, as coded in the IPUMS variable occ1950.
It is reasonable to suppose that among farmers, those who owned farms were positively selected in terms of their underlying skills or productivity. We have little direct evidence on this issue in our linked dataset. For some insight, we have undertaken a supplementary analysis of the 1870 Census of Population data (Ruggles et al. 2021), which reports the value of real estate assets. We restricted the sample to Black farmers who were household heads and had children, and we assumed that positive values of real estate wealth corresponded to farm ownership. In this sample, there is some evidence of positive selection in that Black farmers who owned property had higher literacy rates than those who did not, even when controlling for age and region of residence (by about 12 percentage points). This finding reflects ownership patterns only five years after the end of the Civil War, and we caution that much may have changed between 1870 and 1880.
Further evidence on farm size and value in 1880 is contained in Figure 1. The upper panels focus on each farm’s total acreage, including improved and unimproved acreage. Here, we assign farmers who rent land a value of zero acres in the histograms. Only 17 percent of Black farmers owned and operated farms with 40 acres or more, and only 11 percent owned 80 acres or more. Conditional on owning land, the average total acreage for Black farmers was 101 acres (median 66). In contrast, the majority of White farmers in our sample owned land, and 73 percent nationwide (70 percent in the South) operated farms with 40 acres of land or more.Footnote 27 Approximately 58 percent of White farming fathers owned and operated 80 or more acres of land nationally (56 percent in the South), five times the rate among Black farmers. Conditional on owning land, White farmers operated farms with an average of 180 total acres (median 120).
1880 DISTRIBUTION OF OWNED FARM ACREAGE AND FARM VALUE AMONG BLACK AND WHITE FARMERS, ALL U.S. STATES
Notes: Includes non-owning farmers but not farm laborers. For presentation, farm acreage capped at 300 acres, and farm values are capped at $2,000. All renters for share are assumed to have a farm value of $0. Renters for cash are assumed to have a farm value equal to their livestock value only. Histograms are weighted plots of 4,934 White farmers and 3,182 Black farmers.
Sources: Authors’ collection and transcription of manuscripts from the 1880 Census of Agriculture, as described in the text and Online Appendix.
Summary statistics for farm values (expressed in 1880 dollars, as reported in the census manuscripts) reveal similar racial inequalities, as evidenced in the lower panels of Figure 1. The vast majority of Black farmers—89 percent—owned properties valued at $500 or less, compared to only 27 percent of White farmers. This reflects, in part, the large number of Black farmers who owned no land at all (hence the spike at $0). At the upper end of the distribution, only 2 percent of Black farmers operated farms valued above $2,000, compared to 43 percent of White farmers nationally and 19 percent of White farmers in the South. Even among the farm owners, the average value of Black farms was low compared to White farms.
Large disparities in farm ownership rates, acreage, and values in 1880 are not a great surprise.Footnote 28 And yet the histograms show in sharp relief the strikingly wide Black-White gaps in landholding patterns at the end of Reconstruction. In turn, this highlights what was at stake in the debates surrounding land policy at the end of the Civil War and early years of Reconstruction (Cox Reference Cox1958; Oubre 1978; Foner 1988), when recently emancipated people held out hope for, in Foner’s words, “a modicum of economic independence” and “land of their own” (p. 104).
Looking forward to the next generation and beyond, it is impossible to know what course the political economy of the South and the fortunes of Black Americans’ might have taken if there had been large-scale land re-allocations after the Civil War (Du Bois 1935, p. 602; Ransom Reference Ransom2005). Such a counterfactual extends far beyond the bounds of what comparisons within our sample can reveal and would require, at a minimum, careful modeling of economic and political spillover effects. But it is possible to see in our sample how the children of Black farmers fared later in life as a function of their parents’ landholding in the economic environment that actually did prevail in the late nineteenth century. In this setting, if selection of fathers into landownership was positive, as suggested previously, then evidence of positive status transmission may be considered an upper bound relative to an experiment with good-as-random assignments of land, all else being the same.
Sons of Farmers: Intergenerational Results
The panels on the left side of Figure 2 report the outcomes of farmers’ sons observed in 1900. On the x-axis, we classify 1880 households into six groups based on total owned acreage or (separately) farm value: those with “zero” and then five quintiles for those with at least some owned acreage or value. Then, we plot the average value of sons’ outcomes for each group, with separate lines corresponding to the acreage- or value-based quintiles.Footnote 29
SONS’ 1900 OUTCOMES BY FATHERS’ 1880 LANDOWNING OR OCCUPATION STATUS
Notes: Figures display the average value for each outcome within each quintile or occupation category after age adjusting. The horizontal reference lines are average outcomes for White sons, as described in the legends. For the Quintiles of Farm Ownership panel, the reference sample for age adjustment is all farmers (owners, cash renters, and share renters). For the Occupation Categories panel, the reference sample for age adjustment is all fathers. Reference samples are divided into age bins of 18–24, 25–29, 30–34, and 35–39. Quintile-by-race and occupation-by-race subsamples are then reweighted to reflect the age distribution of the reference sample. Distributions in left panel (by value and acreage) reflect 4,934 White farmers and 3,182 Black farmers. Distributions in right panel (by occupation) reflect 282,938 White father/son pairs and 25,025 Black father/son pairs.
Sources: See the text and Online Appendix.
It is clear that some advantages accrued to Black sons of fathers who owned larger or more valuable farms. They tended to have higher literacy and homeownership rates and were, for the most part, less likely to be laborers (top three panels).Footnote 30 These patterns were not monotonic across categories on the x-axis, but there are usually clear differences between sons from the top quintiles and those from the lowest ranking category. For instance, Black sons from the top quintile of Black farming families (measured as either value or acreage) had a 75 percent chance of being literate in 1900 and a 15 to 20 percent chance of being homeowners. For comparison, Black sons of farmers with no owned acreage or farm value were about 60 percent literate, and only 8 percent owned homes in 1900. In terms of the income score rank variable, the intergenerational returns associated with landholdings were less pronounced (bottom panel).
Despite some advantages for Black sons whose fathers owned larger or more valuable farms, these were small compared to the magnitude of the Black-White gaps. For reference, we have plotted two horizontal lines in each panel showing the average outcome for White sons whose fathers owned no land, both nationwide (dashed line) and within the South only (solid line). For additional detail, results for White sons by fathers’ quintiles of farm value and acreage are contained in Online Appendix Figures A2 and A3.
It is striking that Black sons from the top quintiles of farming families (by either acreage or value) were far less likely to be literate than White sons from the bottom—by about 20 percentage points—and were substantially more likely to work as laborers in 1900. Of course, these gaps are even larger when comparing sons of landless Black farmers to those of landless White farmers. Only in terms of homeownership did Black sons from the best-off farming families approach parity with sons of landless White farmers. Overall, on each of the four metrics, Black farmers’ sons from every quintile fared worse than sons of White farmers who did not own land. To see this disparity another way, we pooled the Black and White samples and limited the analysis to sons of farmers holding no more than the 99th percentile of Black landholdings in 1880 (320 acres). Linear regressions of sons’ outcomes on fathers’ landholdings, an indicator for race, and the interaction of race and landholdings confirm large differences in the intercepts for Black and White sons, whereas the slopes are rarely statistically different.Footnote 31
For reference and robustness, we summarize the baseline results using regressions with additional control variables in Online Appendix Table A1. The regressions include fixed effects for the age of the father (1880) and son, state of residence in 1880, and residence in an urban location in 1880. The results are consistent with those graphed in Figure 2. Although differences between adjacent quintiles are often small, the general pattern of children from households in the top quintiles faring better than those from the bottom is robust, particularly for measures of homeownership, literacy, and laborer status. We report regression estimates for several additional outcomes, including migration, marital status, professional status, and family size, in Online Appendix Table A2.
All Sons: Intergenerational Results
For a wider scope on intergenerational mobility patterns, we can add the sons of non-farmers to the sample and then measure differences in sons’ outcomes depending on their fathers’ occupational group. The fathers’ occupation categories correspond to those in Table 1, but for concise presentation, we combine skilled and semi-skilled blue-collar fathers into a single “blue collar” group. The panels on the right side of Figure 2 plot age-adjusted average outcomes for sons from each category of fathers’ occupations; for reference, the dashed line represents the average outcome for sons of White laborers.
Black sons of land-owning farmers were much more likely to own homes than were Black sons of laborers or any other occupational group. This is important in light of the idea that facilitating more access to landownership for the formerly enslaved population might have translated into economic advantages for the next generation. Moreover, Black sons of white-collar workers, blue-collar workers, and farm owners were all much more likely to be literate than sons of laborers, by at least 10 percentage points and by more than 20 percentage points for sons of white-collar workers.Footnote 32 Sons of laborers, share renters, and cash renters had roughly similar and relatively low literacy rates. This is consistent with the results for farm families discussed previously, where farm owners’ sons were more literate than renters’ sons. It also confirms the sense that children from better-off Black families had substantial advantages relative to other Black children in the acquisition of human capital.
Sons of unskilled laborers were substantially more likely to be unskilled laborers themselves than were the sons from all other groups, which were roughly similar to one another in their likelihood of being laborers (around 50 percent). In the bottom right panel of Figure 2, the sons of white-collar and blue-collar Black fathers achieved somewhat higher average occupational ranks than other groups, even though their probability of being a laborer was similar.
The striking size of the Black-White gap in sons’ outcomes is again apparent in the right-side panels of Figure 2. In terms of homeownership, Black sons of land-owning farmers were the only group that fared better than White sons of laborers nationally (dashed horizontal line), and they still fell below the “South only” line for White sons of laborers (solid horizontal line). In terms of literacy, even sons of white-collar Black workers were less literate than sons of White laborers nationally, though they fared better than the sons of southern White laborers. In terms of laborer status and occupational income rank in 1900, Black sons from across the fathers’ occupational distribution fared worse than White sons of laborers, even when limiting the comparison to sons of southern White men.
Figure 3 reports several additional outcomes for Black sons according to their fathers’ occupational category. Black sons of white-collar workers were far more likely than others to hold a white-collar occupation in 1900 (top panel), consistent with both their relatively high literacy rate and a tendency for intergenerational persistence in occupations. Black sons of farmers, whether farm owners or renters, had relatively high marriage rates and more children by 1900 (middle panels). Sons of farmers also had relatively low rates of inter-state and inter-regional migration (bottom panels), whereas sons of white- or blue-collar workers had substantially higher rates of migration.
SONS’ 1900 ADDITIONAL OUTCOMES BY FATHERS’ 1880 OCCUPATION CATEGORIES
Notes: See notes to Figure 2. The horizontal dashed line represents the average outcome level for White sons of laborers.
Sources: See the text and Online Appendix.
Results in Light of Other Research on Race and Mobility in U.S. History
Our findings regarding sons’ literacy and its positive relationship with fathers’ landownership or occupational status are broadly consistent with evidence from Miller (Reference Miller2020, table 8). That paper shows that Black children in the Cherokee Nation in Oklahoma—whose parents were entitled to claim land after emancipation—were more likely to attend school than Black children elsewhere. This dimension of outcomes is important because racial gaps in literacy and schooling were enormous in the immediate wake of emancipation. The gap in literacy closed over time as each new generation of Black students improved upon the previous generation’s attainment. Yet this might not have been sufficient to narrow the Black-White gap in “years of education” for late nineteenth-century birth cohorts, due to segregated schooling and differences in school quality and access to higher levels of education (Collins and Margo Reference Collins, Margo, Hanushek and Welch2006; Carruthers and Wanamaker Reference Carruthers and Wanamaker2017). Building on Margo (1990, 2016) and new data resources, we would encourage more micro-level research on this era to better understand the opportunities for (and barriers to) Black families’ acquisition and transmission of human capital following emancipation.
The results in this paper are also broadly consistent with those in Collins and Wanamaker (Reference Collins and Wanamaker2022). In both this paper and Collins and Wanamaker (Reference Collins and Wanamaker2022), White sons fared much better than Black sons, conditional on their fathers’ initial status. In this sense, there was a large and empirically important “mobility gap” between the Black and White populations that propagated large “initial differences” in economic status. Moreover, White sons from low-ranked households fared better on average than Black sons from high-ranked households, underscoring the magnitude of the differences in lifetime opportunities for Black and White children. After assigning occupational income scores and ranking fathers and sons in 1880 and 1900, Collins and Wanamaker (Reference Collins and Wanamaker2022) show that if relatively poor Black children had been able to move up the occupational income distribution to the same extent that poor White children did, the resulting overall Black-White gap in labor outcomes in 1900 would have been far smaller.
This paper’s results are also broadly consistent with those reported in Jácome, Kuziemko, and Naidu (Reference Jácome, Kuziemko and Naidu2022, figure 3), which begins with a birth cohort from the 1910s and assigns income scores to fathers and children. That paper also finds large disparities between Black and White children’s occupational scores as adults, conditional on their fathers’ estimated occupational status.Footnote 33 This is reassuring since their paper is based on a different kind of dataset than ours—retrospective surveys rather than linked census data.
EVIDENCE ON CHANNELS AND HETEROGENEITY
As pointed out earlier, intergenerational patterns of labor market outcomes may reflect investments in human capital, social connections, local resources and environments, and more (Loury Reference Loury1977; Becker and Tomes Reference Becker and Tomes1986; Solon Reference Solon, Ashenfelter and Card1999; Margo 2016). In the context we study, how were landowning Black families able to endow their sons with higher homeownership and literacy rates in this first post-Civil War generation? A deep assessment of this question is beyond the scope of this paper, but we can shed some light on broad, proximate channels.
It is possible, in theory, that location-specific differences in economic opportunity explain both land ownership rates among Black fathers and the subsequent achievements of sons. For a simple test of whether the father-son gradients in outcomes described in the previous section are largely a reflection of differences in local environments, we can regress sons’ outcomes on fathers’ status conditional on location fixed effects. We implement this test using indicators for “state economic areas” (SEAs), groups of contiguous counties that had similar economic characteristics in the twentieth century (Bogue 1951).Footnote 34 If cross-place differences were driving the baseline results described earlier, we would expect the relationship between fathers’ and sons’ outcomes to be much weaker when based solely on within locality variation.
The results are reported in Column (2) of Table 2 (sons of farmers) and Table 3 (all sons). In sum, we find that sons’ homeownership and literacy advantages associated with fathers’ farm ownership are not much weaker with the addition of location fixed effects. Therefore, we infer that the mechanisms linking fathers’ and sons’ outcomes were likely to have been primarily “within family” in nature.
RETURNS TO BLACK FATHERS’ FARM VALUE
Notes: The sample includes sons of farmers only. The table contains regression coefficients on quintile dummies, representing the quintile of the value of fathers’ 1880 farm holdings. Coefficients are interpreted relative to a farm value of $0. Fathers who owned farms are assigned the combined value of the property and livestock. Fathers who rented for cash are assigned a value equal to the value of farm livestock. Fathers who rented for a share are assigned a value of $0. Baseline regressions include fixed effects for sons’ age in 1900, fathers’ age in 1880, fathers’ urban status, and fathers’ 1880 state of residence. Column (2) includes fixed effects for State Economic Area (SEA) of residence in lieu of state of residence fixed effects. Column (3) includes only the oldest sons in the 1880 household. Columns (4) and (5) bifurcate the sample by age. Column (6) excludes Black sons who were farmers or farm laborers in 1900.
Sources: See the text and Online Appendix.
RETURNS TO BLACK FATHERS’ OCCUPATION
Notes: See notes to Table 2.
Sources: See the text and Online Appendix.
Further analysis of heterogeneity within the linked sample provides more nuance. Across Columns (4) and (5) in Tables 2 and 3, we split the sample by sons’ ages (less than 30 versus 30 and over). We see that fathers’ landholdings are similarly correlated with sons’ literacy in both subsamples, but the father-son property ownership gradient is stronger among sons observed at older ages. Thus, observable advantages in homeownership, a proxy for household wealth, tended to materialize later in the sons’ lifecycle. It is tempting to infer that this simply reflects within-family transfers or inheritances of land (e.g., sons taking over the family farm). Surely there was some of this, but the story is more complex. We see a positive, albeit muted, father-son gradient in ownership even among sons who left agriculture (Column (6)), consistent with improved lifecycle wealth accumulation for Black sons of landowning farmers even if they left the family farm.Footnote 35 In addition, we find no consistent evidence of a stronger gradient in home ownership for the oldest sons in each household (Column (3)).
Why were Black sons of landholding fathers unable to translate their advantages in terms of family background and literacy into much better labor market outcomes, at least by the metrics we observe? We offer three interconnected explanations. First, it is important to keep in mind that most landowning Black households were still relatively poor—they were higher on the “agricultural ladder” than many others and, therefore, had a higher degree of autonomy in their work and a higher average income, but their economic resources were still modest. Second, pervasive discrimination impeded Black advancement in education as well as up the occupational ladder into high-paying positions. This does not imply that labor market returns to literacy or family background were near zero circa 1900, but there were limited possibilities for earning high incomes through upward occupational mobility. Third, fathers’ property ownership and place-specific human capital may have entailed substantial returns for sons within farming—perhaps in ways that our measures cannot detect—but not elsewhere. If so, sons of landowning fathers may have been strongly attached to local agriculture and less inclined to seek opportunity in other lines of work. This is consistent with evidence of their low rates of inter-state and inter-regional migration compared to sons of non-farming fathers.Footnote 36
CONCLUSIONS
In 1880, 15 years after the Civil War’s last battle, there was substantial heterogeneity in the economic situation of Black households. Some had succeeded in acquiring land of their own, but most had not; some held skilled-craft or white-collar occupations, but most toiled as laborers or sharecroppers. By combining information from the agricultural and population census records, we provide a fuller picture of this heterogeneity than was previously available, as well as a clearer view of the wide gap between the Black and White distributions.
We then show that Black households with more economic resources were able to transmit substantial advantages to the next generation, particularly in terms of literacy and property ownership. We also find that this positive transmission did not merely reflect better local conditions; rather, the relationship appears to have held across families within the same 1880 locality. This suggests that within-family mechanisms were largely responsible for the transmission of status, though we would also argue that more research along these lines is merited to better understand both within-family and across-place factors.
These advantages, however, were quite limited when seen in the overall landscape of racial disparities. When we look specifically at the labor market outcomes of young men in our linked dataset, we see only traces of advantages transmitted over generations within the Black population, and we see large Black-White gaps in 1900 regardless of the sons’ background. It is striking that the children of White farmers who did not own land or of unskilled laborers typically had better adult outcomes than did the children of Black workers in any occupational category. These patterns underpin a fact emphasized in Margo (2016): Black-White income convergence after 1870 was far slower than standard estimates of intergenerational status persistence would predict.
Vast differences in Black and White landholding, wealth, and opportunities for economic advancement were well entrenched aspects of American society circa 1880. Ransom points out that by this time, “most Northerners were unwilling to pay for the effort that would be required to guarantee freed slaves their rights in a society of hostile white Southerners” (2005, p. 367). Without these protections, even the children of landholding Black households had difficulty advancing and narrowing gaps in human capital and economic outcomes relative to White children, ensuring the perpetuation of large disparities into the twentieth century and beyond.
'/%3e%3cpath%20d='M17%207.971L15.711%206.677%2010.912%2011.495%2010.912%200%209.088%200%209.088%2011.495%204.289%206.677%203%207.971%2010%2015z'%20transform='translate(8%208)'/%3e%3c/g%3e%3c/g%3e%3c/svg%3e)
