The Confederate Diaspora

The Confederate Diaspora Abstract This paper shows how white migration out of the postbellum South diffused and entrenched Confederate culture across the United States at a critical juncture of westward expansion and postwar reconciliation. These migrants laid the groundwork for Confederate symbols and racial norms to become pervasive nationally in the early 20th century. Occupying positions of authority, former slaveholders played an outsized role in this process. Beyond memorializing the Confederacy, migrants exacerbated racial violence, boosted novel forms of exclusion, and compounded Black disadvantage outside the South. Moving West, former Confederates had larger effects in frontier communities lacking established culture and institutions. Over time, they continued to transmit norms to their children and non-Southern neighbors. The diaspora legacy persists over the long run, shaping racial inequities in labor, housing, and policing. Together, our findings offer a new perspective on migration, elite influence, and the interplay between culture and institutions in the 1 Introduction The history of the United States is marked by racial division and conflict. The Civil War (1861–65) brought about the end of slavery, while the civil rights movement of the 1960s dismantled most de jure forms of racial discrimination. Despite this overall trend of racially-progressive reform, stark disparities persist in critical areas such as education, housing, labor markets, and policing. Moreover, monuments and iconography venerating the pro-slavery Confederacy, which became widespread across the nation in the 20th century, still dot the American cultural landscape. These symbols of past injustice, coupled with persistent inequities, raise recurring questions about the legacy of slavery and the extent of racial animus entrenched within local institutions, civil society organizations, and norms.1 The enduring influence of the Confederacy, despite its resounding military defeat, owes in part to the “Lost Cause of the Confederacy,” a revisionist narrative that emerged in the postbellum U.S. South (Domby, 2020). Seeking to redeem the image of the South, Lost Cause ideologues offered noble rationalizations for Southern secession and the war, downplaying slavery while emphasizing Northern aggression and the defense of states’ rights (Cox, 2003). White supremacist tropes were central to this retelling, touting the supposed contentment of enslaved people and Black inferiority as just cause for slavery (Blight, 2001; Cowa, 2013). Advocates, such as the United Daughters of the Confederacy (UDC), propagated this narrative across the country, through various forms of Confederate memorialization (Chamberlain and Yanus, 2021; Cox, 2021; Waite, 2020). At the same time, more overt and violent expressions of racial animus gained momentum as the Ku Klux Klan (KKK) and lynchings spread. Together, these actions coalesced into a budding “Confederate culture,” which characterized the postbellum South and, in the early 20th century, transformed America’s collective memory of the Confederacy.2 This paper investigates the role of Southern white migration after the Civil War in diffusing Confederate culture throughout the country at a critical juncture of nation building. Estimates based on linked Census records suggest that nearly one million whites—including more than 61,000 former enslavers and 127,000 of their household kin—left the former Confederacy for the rest of the U.S. in the three decades after the war. These migrants and their descendants retained strong attachments to Confederate culture; e.g., like their brethren in the South, they gave their newborn children the names of prominent Confederate leaders long after the war. We explore how this “Confederate diaspora” influenced the cultural landscape and, in the process, changed the trajectory of racial inequity across America.3 We begin by characterizing the formation of the Confederate diaspora, tracking migrants, including former slaveholders, and their occupational choices outside the South from 1870 to 1900. We show that these migrants often went west, settling in nascent communities throughout the frontier where institutions were relatively weak and cultural norms not yet deeply ingrained. At destination, Southern migrants sorted into positions of authority—judges, lawyers, public administration, police, and clergy— much more so than other residents, including migrants from elsewhere. Such choices went beyond a 1 Recent polling from Morning Consult (2021) found that 51% of respondents and 60% of whites favored keeping Confederate leader statues while 30% and 25% favored removing them, respectively. Similarly, 47% of respondents and 54% of whites viewed the Confederate battle flag as a symbol of Southern pride, versus 36% and 30% seeing it as a symbol of racism. 2 Historian Karen L. Cox (2003, p. 1) defines “Confederate culture” as “those ideas and symbols that Lost Cause devotees associated with the former Confederacy.” Historian Kevin Waite (2020), meanwhile, uses “Confederate culture” to describe an affinity for Confederate memorialization in many places outside of the South during the 20th century. 3 We describe early postbellum migrants as the “Confederate diaspora,” distinguishing it from the later “Southern diaspora,” comprised of those who left the South during the Great Migration of the 20th century. We use the latter term, inspired by Gregory (2005), in a complementary study detailed below (Bazzi et al., forthcoming). 1 broader sorting into high-status occupations and was especially strong for migrants from slaveholding households in the antebellum South, who often assumed leadership roles in newly incorporated areas. Next, we show how the diaspora brought symbolic and material expressions of Confederate culture throughout the country by the early 20th century. We focus on four outcomes: (i) Confederate memorialization (e.g., monuments, place names), (ii) UDC chapters, (iii) KKK chapters, and (iv) lynchings of Black people. Together, these measures capture a process, described by historians of the postbellum U.S., in which Southern whites mobilized grievances and engaged in racial terror in order to recreate the social and economic hierarchies of the old South. We aim to identify a causal impact of the diaspora in catalyzing this process beyond the South. We develop a shift-share instrumental variable (SSIV) framework to isolate exogenous county-level variation in exposure to postbellum Southern white migration through 1900. Our SSIV combines historical migrant networks outside the South as of 1870, at the dawn of the postbellum period (shares), with predicted migration flows out of the South from 1870 to 1900 (shifts). Conditional on the population share of Southern-born whites in 1870, our SSIV estimates identify the distinct influence of the Confederate diaspora forged between 1870 and 1900. We find significant cultural impacts of the Confederate diaspora in destination counties. These migrants helped build the physical infrastructure (memorials) and organizational foundations (UDC) for keeping the memory of the Confederacy alive. By hastening the diffusion of the KKK and Black lynchings, they helped lay the groundwork for deep racial injustice in public life. Our findings are robust to (i) a large set of flexible controls capturing migrant sorting correlates, (ii) inference corrections for SSIVs and spatial regressions, (iii) omitting individual origin and destination states, (iv) excluding border states outside the Confederacy where enslavement was also legal before the war, and (v) restricting to newly incorporated counties with little to no population in 1860. The latter provides a strong check against biases due to endogenous sorting towards places with a longstanding, ideologically-aligned antebellum population. Together, these results, and other SSIV diagnostics, point to a causal interpretation. Our IV estimates are consistent with sizable impacts of a relatively small Confederate diaspora. On average, migrants from the former Confederacy comprised 2.2% of the population in destination counties by 1900. Yet, their influence appears outsized. For example, moving from zero to the mean Confederate diaspora implies an 8 percentage point (p.p.) increase in the likelihood of KKK activity relative to a mean prevalence in 35% of counties outside the South. The effects on post-1900s lynching events were even larger, growing in likelihood by 4 p.p. relative to a mean of 5%. These large effect sizes suggest that the reach of these migrants may have extended well beyond the diaspora itself. Beyond their immediate impacts on the diffusion of Confederate culture, the diaspora also contributed to novel forms of racial exclusion outside the South, namely, the “sundown town.” Distinct from de jure exclusionary institutions associated with the Confederacy and Jim Crow, sundown towns spread throughout the non-South from 1890–1960. Historical research by Loewen (2005) describes the process by which Blacks and other minorities were excluded from residing in these all-white towns after sunset. We link Southern white migrants to the prevalence of sundown towns and show, at the town level, that a 1 p.p. increase in the size of the Confederate diaspora in 1900 led to a 2.4 p.p. increase in the likelihood of Black depopulation, i.e., towns with ≥ 25 Black residents in 1870 having zero Blacks after 1900. This sort of racial cleansing changed the geography of Black settlement throughout the 20th century and, by limiting intergroup contact, may have made it easier for Confederate norms to persist. These early impacts of the Confederate diaspora are not a mere artifact of history but rather continue 2 to cast a long shadow over the socioeconomic well-being of Black Americans. Using our county-level SSIV strategy, we find that a larger diaspora in 1900 is associated with larger racial wage gaps, greater residential segregation, and higher rates of Black incarceration and police-induced mortality through the 21st century. Such disparities are mirrored in more conservative racial attitudes among whites in these counties, which we identify using modern survey data. Together, our findings suggest that, despite racially progressive national policy reform since the 1960s, inequity persists locally through institutionalized mechanisms reinforced by a culture of racial animus. We provide rich and varied evidence on the process by which the diaspora shaped these institutional and cultural foundations of racial inequity. First, we show that migrants more effectively diffused Confederate culture in communities with more malleable social and political infrastructure. In particular, the diaspora impacts were significantly larger in counties along the frontier (as measured in Bazzi et al., 2020), where state institutions were weak and where culture was incipient. At the same time, the diaspora transmitted Confederate culture more effectively in counties with fewer Union Army enlistees whose families after the war might oppose such resurgent ideology in public life. Second, we find that former slaveholders, despite being a minority of Southern white migrants from 1870 to 1900, became pivotal to the diffusion and entrenchment of Confederate culture. We identify their sizable influence using an analogous SSIV for enslavers, which, conditional on the instrumented overall diaspora, relies on distinct identifying variation. These former elites from the South not only imported Confederate norms but plausibly helped to embed them in nascent institutions. From their outsized presence in local public administration and religious leadership, they amplified memorialization and facilitated racial terror. Such influence was not merely symbolic: the likelihood of incarceration in 1920 was twice as large for Black men who, in 1900, lived in counties with migrants from slaveholding backgrounds working in policing, judicial, or administrative occupations. Third, we identify overrepresentation of first- and second-generation migrants in the KKK. Established in the South in 1915, the second KKK helped to rejuvenate and mainstream Confederate culture. Using newly digitized membership records, we show that, relative to men with no Southern heritage, those born in the South were 11% more likely to belong to the KKK in the Denver metropolitan area, a major hub of Klan activity outside the South in the 1920s. We find a similar differential for secondgeneration migrants born in the diaspora, and these patterns hold within Census enumeration blocks. This suggests that vertical cultural transmission within families may have been an important vehicle for sustaining diaspora influence long after the initial Confederate migrants had passed. Fourth, we provide suggestive evidence of cultural spillovers as migrants interacted with their nonSouthern neighbors. White men without Southern heritage living next door to either first- or secondgeneration Southern whites were significantly more likely to join the KKK in Denver. Together with the higher rates of KKK membership among Southern whites, these patterns are consistent with cultural transmission from Southern to non-Southern whites. We offer further evidence on the direction of transmission using another measure of Confederate culture—whether a child is given a name associated with prominent Confederate leaders—available for the entire non-South. After moving to counties with a larger Confederate diaspora, non-Southern white parents gave their later-born children names that are more evocative of Confederate heroes than those given to their children born prior to moving. Finally, we highlight occupational choices as another potentially important channel for sustaining the early influence of the diaspora across successive generations. Much like their parents born in the South, 3 those in the second generation born outside the South are significantly more likely than those without Southern heritage to work in authority positions in the criminal justice system and religious leadership. Moreover, unlike the initial movers, these second-generation migrants are overrepresented in the education sector where they could have cultivated ideology during students’ impressionable years. Overall, this intergenerational transmission of authority roles may have helped Confederate norms endure. This paper offers new insights on the interplay between culture and institutions. A large and influential literature characterizes the origins and consequences of institutions (e.g., Acemoglu et al., 2005; North et al., 2009) and culture (e.g., Tabellini, 2010). The links between these two key forces in comparative development, until recently under-explored, are the focus of an emerging body of research (see Tabellini, 2008; Alesina and Giuliano, 2015; Bisin and Verdier, 2023). Understanding how culture and institutions influence each other can be challenging empirically. Using granular data on migration, occupational choices, and local governance, we shed unique light on the historical process by which Confederate cultural entrepreneurs captured early institutions across America. These deep roots may explain some of the limited progress towards racial justice in the U.S.: laws to reduce racial discrimination may have limited impact where a culture of racial animus is ingrained in local institutions. While we emphasize an institutional interpretation, our findings are also consistent with prestigebiased cultural transmission wherein the non-elite masses seek to emulate former slaveowners in positions of power, perhaps confounding the latter’s white supremacist ideology with their socioeconomic success. Henrich and Gil-White (2001) formalize this type of evolutionary transmission process in which cultural change effectively follows salient leaders. Along these lines, Dippel and Heblich (2021) show that a small but politically active German immigrant population formed a vanguard of anti-slavery mobilization in the mid-20th century: they swayed white Americans to join the Union army during the Civil War and changed the trajectory of local support for racial progressivism over the long run. The centrality of former slaveowners in the Confederate diaspora legacy points to analogous influence.4 Our paper illuminates a hidden element in the history of nation building in the early 20th century United States. This was a time of national reconciliation between former Confederate and Union territories as Southerners, Northerners, and Westerners began to forge new shared identities in the postbellum country. The bridging of the North–South divide during this period is well documented in the seminal history by Blight (2001). Our findings establish the critical and understudied role of the Confederate diaspora in this process. These migrants brought distinctive cultural norms and experience with slaveholding to new areas of the country. Their location and occupational choices sowed the seeds of Confederate affinity and racial chauvinism that later erupted across white America after the wildly popular 1915 film, The Birth of a Nation (Ang, 2023; Esposito et al., 2023).5 More specifically, we offer a new perspective on the deep roots of racial animus across the U.S.6 A burgeoning literature in economics explores the consequences of racial bias and discrimination across 4 Our findings on the sociopolitical influence of slaveowners outside the South resonate with research on the recovery of slaveholder power within the postbellum South (see, among others, Foner (2002) on the undermining of racial progress during the Reconstruction era and Ager et al. (2021) on intergenerational mobility within former slaveholding families). Beyond our setting, these findings mirror elite influence observed among migrating Nazis after WWII (Ochsner and Roesel, 2020) and Luther-adjacent local leaders during the Protestant Reformation (Becker et al., 2020). 5 Perhaps the best, albeit extreme, example of this is President Woodrow Wilson. Born in Virginia in 1856, Wilson settled in New England during the 1880s, later becoming president of Princeton University, governor of New Jersey, and, in 1913, the president of the United States. During his tenure, The Birth of a Nation was the first film to be screened at the White House. 6 Such work spans economics (e.g., Aaronson et al., 2021; Fryer Jr. and Levitt, 2012; Shertzer and Walsh, 2019), history (e.g., Campney, 2019), political science (e.g., Acharya et al., 2016), and sociology (e.g., O’Connell, 2019), among others. 4 numerous domains, including labor markets (e.g., Bayer and Charles, 2018), media (e.g., MorenoMedina et al., 2022), credit (e.g., Bayer et al., 2018), education (e.g., Billings et al., 2014), housing (e.g. Logan and Parman, 2017), public accommodations (e.g., Cook et al., 2022), policing (e.g., Knox et al., 2020), and the criminal justice system (e.g., Arnold et al., 2022). Our research provides fresh evidence on the cultural foundations of institutionalized animus that underlie some of these proximate causes of racial inequity. Much progress has been made in uncovering such foundations within the South, including work connecting (post-)Confederate institutions to adverse outcomes for Black Americans.7 Our paper is among the first to show how these institutions echoed across the country through internal migration in the decades after the Civil War. Together with Bazzi et al. (forthcoming), we break new ground in the study of Southern migration and its influence on cultural and political change. Bazzi et al. (forthcoming) show how mass Southern white migration in the 20th century changed the trajectory of partisan politics by creating new possibilities for a broad-based national coalition of religious, racial, and economic conservatives. By contrast, the present study provides a micro lens on the distinct influence of the smaller Confederate diaspora, and especially the slaveholder class, as it brought extreme ideologies to new parts of America at a critical juncture of westward expansion in the late 19th century. These early migrants mobilized postwar grievances, captured local institutions, and built some of the networks that would shape the legacy of the Great Migration of whites and Blacks in the decades to come. Our empirical evidence complements a rich historical literature on Southern emigration (Berry, 2000; Dippel, 2005; Dochuk, 2010; Gregory, 2005; Waite, 2021), while offering a new angle on the evolution of racial hierarchies. The enduring legacy of the early postbellum diaspora is a testament to the cultural influence of small numbers of immigrants in the early stages of development, a common theme across the American landscape with important political implications today (Bazzi et al., 2020; Grosjean, 2014; Zelinsky, 1973). Before proceeding, we offer a note on interpretation. Our findings show that, on average, Confederate migrants diffused racial animus outside the South. Our results do not imply that all migrants were equally attached to Confederate culture; they varied in ideology and influence, with former enslavers playing an especially important role. Moreover, our results do not imply that these migrants were alone in propagating animus; an important factor in the spread of Confederate culture was the collaboration or resistance from non-Southern whites—who also contributed to animus and exclusion in other ways. The paper proceeds as follows. Section 2 provides background on the Confederate diaspora and describes some of our data contributions. Section 3 characterizes the location and occupation choices of migrants. Section 4 presents our main county-level estimates, showing how postbellum Southern white migration diffused Confederate culture and racial norms throughout America. Section 5 looks beyond initial migrants, to identify channels through which they catalyzed cultural and institutional change. Section 6 studies the long-run legacy of the diaspora on racial inequity. Section 7 concludes. 7 See, among others, O’Connell (2020) and Williams (2021) on memorialization and wage inequality, Henderson et al. (2021) on memorialization and lynching, Rahnama (2022) on de-memorialization and attitude change, Jones et al. (2017) and Williams (forthcoming) on lynching and voting, and Cook (2014) on lynching and patenting. There is also a much broader literature tackling the various legacies of slavery (see, e.g., Althoff and Reichardt, 2022; Bertocchi and Dimico, 2012, 2014; Buttrick and Mazen, 2022; Cook et al., 2018; Jung, 2019; McVeigh et al., 2014; Nunn, 2008; O’Connell, 2012; Reece and O’Connell, 2016; Suryanarayan and White, 2021). 5 2 Historical Background and Data This section details the history of Southern white outmigration from the antebellum era through the postbellum era up to 1900. We also trace the configuration of “Confederate culture,” from Lost Cause narratives and public memorialization to expressions of white supremacy in the South and, later, across America. Besides providing background, we describe the rich primary and secondary data that we use to track the diaspora and to measure its cultural and institutional impacts outside the South. 2.1 The Confederate Diaspora In the mid-19th century, the Gold Rush ushered in the first major waves of Southern white outmigration. Poor settlers pushed westward in pursuit of cheap land in places where slavery—and the labor competition it entailed—had not yet been established. At the same time, many wealthy slaveholders were keen to expand plantation agriculture and to bring slavery to fertile areas of the West (Waite, 2021). Despite their differences, early Southern white migrants were rather unified in their commitment to preserving racial hierarchies and white supremacy (Dippel, 2005). One such early migrant was Cameron E. Thom, whose story, retold by Waite (2021), illustrates the journey of slaveholding elites in the Confederate diaspora, their sorting into authority positions, and their role in mobilizing influential organizations like the KKK and the UDC. A Virginia-born lawyer, Thom first went west in 1849 with several slaves in pursuit of gold. He settled in Sacramento, where he opened a law office, but left to serve as a Confederate captain in the Civil War. Returning to California after the war, he oversaw, as the district attorney of Los Angeles, a sham trial that freed white defendants in the Chinatown massacre of 1871. He later became mayor of Los Angeles in 1882 and went on to co-found the town of Glendale in 1889, which emerged as a regional and national hub for white supremacy. An early “sundown town,” Glendale hosted one of the first chapters of the UDC outside of the South—cofounded by Thom’s wife, Belle, in 1899—and later retained an active KKK presence. After the Civil War, enslaved people were freed, and plantation agriculture collapsed throughout the South. Combined with wartime destruction, ensuing losses of labor, capital, and credit devastated the Southern economy, including its now-former slaveholder class (Aldrich, 1973; Baker and Hahn, 2016). The sudden emancipation and enfranchisement of Black people threatened to undermine the white monopoly on economic and political power in the region (Acharya et al., 2016). These changes, alongside frustration with Union Army occupation and the advent of Reconstruction, set the stage for new waves of Southern white migration. From 1870–1900, nearly 18% of Southern whites left the region for other non-Southern states and territories.8 Migrants hailed primarily from the upper South, including Virginia, Tennessee, and North Carolina, and, later, from parts of Texas, Arkansas, and Oklahoma. These migrants often sorted into places where they might reproduce the socioeconomic hierarchies of the old South. For some, this meant seeking out all-white destinations, where Blacks could pose no economic or political threat (Dippel, 2005; Waite, 2021). Among the former slaveholding elite, the West offered a chance to replicate the “oligarchic principles of the Confederacy” and hierarchical structures of the antebellum South in new locations and industries (Richardson, 2020, p. 85). Many struck off “in search of land on which to begin rebuilding family estates lost during the Civil War” (Dochuk, 2010, p. 8 As described in Appendix C, we estimate that 921,119 whites left Southern sending states between 1870 and 1900, relative to the 5,163,445 whites of Southern origin living in these states as of the complete-count 1870 U.S. Census. 6 7). Some Southern whites tended to migrate to places like the South politically and agroclimatically, as we show empirically in Section 3. At a broad level, this explains the strong pull West, which boasted large-scale farming opportunities and by the end of the 19th century had more in common ideologically with the South than the North (see Richardson, 2020). Their lived experience under the Confederacy, its crushing defeat, and post-war occupation by the Union Army distinguished early postbellum migrants both from their antebellum forerunners and from those leaving the South during the later, Great Migrations of the 1900s. Their direct experience with slavery, antebellum nostalgia, and animosity towards federal intervention would play an important role in the Confederate diaspora as it put down roots across America at a critical juncture of nation building and post-war reconciliation. Southern slaveholding families may have held especially strong grievances, as they were more likely to join the Confederate Army (Hall et al., 2019) and experienced larger losses in wealth after the war (Ager et al., 2021). Tracking Migrants. We track Southern white migrants using the complete-count U.S. Censuses of Population from 1870–1900. Our main county-level measure of the Confederate diaspora includes those born in former Confederate states plus Oklahoma who lived outside the South in 1900. We also make extensive use of linked records from the Census Linking Project (Abramitzky et al., 2020) to track individuals over time and to construct an instrumental variable for the local diaspora in 1900. Within the overall diaspora, we track the subset of migrants hailing from former slaveholding households. In particular, we link white male slaveholders in the 1860 U.S. Census Slave Schedule to the 1860 U.S. Census of Population.9 We then onward link both listed slaveholders and their male co-resident kin in 1860 to future Censuses through 1900. This procedure delivers the first comprehensive measure of former slaveholder migration across postbellum America (see Section 3.1 for further details). 2.2 Confederate Culture Since America’s early history, a cultural divide has separated North and South. The South had an agroclimatic comparative advantage in crops conducive to large-scale plantation labor, while the North did not. More so than other places, the Southern colonies came to be situated institutionally and economically around the plantation (Coclanis, 2000; Engerman and Sokoloff, 2002). Historical settlement patterns also shaped the cultural divide. Grosjean (2014), for instance, shows how Scots-Irish migrants to the Deep South brought a “culture of honor” favoring a higher prevalence of violence. Moreover, whiteness as an identity may have solidified much earlier in the South, which, unlike the North, did not experience the mass arrival of ethnically diverse European immigrants in the 19th century (Roediger, 2006). Religion further deepened the divide, as a Protestant revival movement known as the Second Great Awakening catalyzed abolitionism across the North through the mid-1800s, and two major denominations, the Baptists and Methodists, split over the issue of slavery in the 1840s (McKivigan, 1999). The fault lines grew deeper around the Civil War. In 1861, seven Southern states declared secession from the United Sates and formed the Confederate States of America, soon joined by four other states once the war began.10 In addition, most of the Indian Territory (later Oklahoma) was aligned with the 9 We match on first and last name as well as county, using the ABE algorithm (Abramitzky et al., 2019) with NYSIISstandardized names. We match nearly 64% of name–county combinations to the 1860 Census, corresponding to over 250,000 former slaveholders in 1860. This compares favorably to linking rates in Ager et al. (2021) and Hall et al. (2019). 10 The first seven included Alabama, Florida, Georgia, Louisiana, Mississippi, Texas, Southern Carolina followed by Arkansas, 7 Confederacy through formal treaties.11 Together, these states and territories organized an army and government seeking to maintain legal slavery. Upon their military defeat, the white residents of the former Confederacy sought to rationalize their loss and to redeem the image of the South. They forged a novel set of symbols and cultural norms as well as new organizations in service of their advancement. These symbols and norms clustered around the “Lost Cause of the Confederacy,” a mythology that emerged in the war’s immediate aftermath, drawing from a publication by the director of a prominent Virginia newspaper (Pollard, 1866). Lost Cause narratives glorified Confederate military figures and offered rationalizations for Southern secession and the war that painted the South in a positive light. A key revisionist maneuver was to call into question the very nature of the war, highlighting Northern aggression against “states rights” as the main motive, rather than slavery itself. Racist tropes of Black inferiority and white supremacy permeated these narratives, which portrayed slaveowners as generously paternalistic (Blight, 2001; Cowa, 2013; Cox, 2003). Prominent advocates, such as the UDC, actively sought to disseminate Lost Cause ideology and to transmit it to the next generation. The UDC hastened the spread of Confederate memorialization across the South in the form of monuments, art, and place names commemorating military heroes and the battle flag. Cox (2003, p. 2) notes how the UDC ensured that monuments “were placed in public settings such as courthouse lawns or town squares, where . . . they could be observed by children,” and “successfully placed Confederate flags in nearly every white public school in the South”; these flags “accompanied portraits of Confederate heroes“ and “were significant in transmitting Confederate ideals.” While many memorials did not reference race, the signaling of white supremacist ideology was well understood by observers, and, moreover, these memorials tended to go hand-in-hand with violent expressions of racial animus, such as public lynchings (Williams, 2021). The Ku Klux Klan helped spread such violence, beginning as an insurgent, white supremacist movement immediately after the war. Suppressed by the federal government during Reconstruction, the group re-emerged in Georgia in 1915 (McVeigh, 2009). Together, these ideals and associated iconography coalesced into a novel Confederate cultural form, giving expression to a revitalized regional identity in the postbellum South. By the early 20th century, Confederate memorialization and Lost Cause narratives had spread across America, perhaps helping to foster support for national reconciliation among non-Southerners (Cox, 2003; Gallagher et al., 2000; Richardson, 2004). Such efforts, alongside battles against a common enemy (the late Indian Wars, the Spanish-American War, and WWI), may have hastened integration across the North–South divide. Later, popular films like The Birth of a Nation in 1915, Gone With the Wind in 1939, and Song of the South in 1946 helped to further solidify a Confederate cultural influence within the national consciousness (Gallagher, 2008). Measurement. We build four measures of Confederate culture at the county level: memorialization, UDC chapters, lynching, and KKK chapters. Together, these measures trace a process in which women in the UDC organized symbolic expressions that bolstered Confederate ideology, while men in the KKK unleashed material expressions of racial terror. Public memorials and lynchings provided technologies North Carolina, Tennessee, and Virginia. Kentucky and Missouri declared membership but were never under Confederate control, and West Virginia became a state in 1863 when it separated from Virginia and joined the Union states. 11 Confederate-Native American alliances in the Indian Territory included “treaties of friendship and alliance” with the “Five Civilized Tribes” of the Cherokee, Chickasaw, Choctaw, Creek, and Seminole, as well as with the Comanche, Osage, Quapaw, Seneca, and Shawnee (Bernholz et al., 2022). Separately, the territory of Confederate Arizona, which consisted of parts of modern-day Arizona and New Mexico, was aligned with but not controlled by the Confederacy. 8 to transmit Confederate memory and racial norms long after slavery and the Civil War had ended. We draw on several primary and secondary sources to track these measures of Confederate culture across time and space. First, we build a new, omnibus measure of memorialization, which first includes monuments from the Southern Poverty Law Center (SPLC)’s “Whose Heritage?” database.12 We also identify prominent Confederate leaders appearing on multiple monuments and look for those leaders’ names in (i) places listed in the U.S. Geographic Names Information System (GNIS), (ii) streets enumerated in the U.S. Census Bureau’s TIGER/Line Shapefiles, and (iii) schools listed in the National Center for Education Statistics (NCES) Public Elementary/Secondary School Universe Survey Data. Second, we geolocate chapters of the United Daughters of the Confederacy (UDC) from 1900–20 based on a novel digitization of the organization’s “Minutes of the Annual Meeting.” Third, we use data on local chapters of the Second KKK from 1915–40 from the Virginia Commonwealth Library’s Klan Map Project. Fourth, we track lynchings of Black people (and other minorities) from 1882–1941 using data from Seguin and Rigby (2019) and the Historic American Lynching (HAL) Project. Migrants in the diaspora often retained Confederate culture. The UDC and the KKK originated in the South and spread outward. Membership in the former was restricted to those with Southern heritage, and hence provides clear evidence of cultural attachments in the diaspora. Memorials arguably do as well. At the individual level, such attachments are evident in KKK membership (see Section 5.3) and in children’s name choices. Figure 2 plots the evolution of Confederate leaders’ names across birth cohorts for three white populations: (i) Southerners in former Confederate states and territories (the South), (ii) Southerners outside the South, and (iii) non-Southerners outside the South.13 We take a conservative approach and consider a child’s first name as belonging to a Confederate leader when containing all of its distinctive constituent parts. This includes, for example, “Robert Lee” but not “Robert” or “Lee” alone, idiosyncratic surnames like “Beauregard,” and unique nicknames like “Stonewall.” Although rare, such names offer a unique signal of attachment to Confederate culture in the historical era. Figure 2 shows that, during the Civil War, Confederate leader names surged in popularity among Southern-heritage whites living within and outside the South. Within the diaspora, this surge was even more pronounced among children born to former slaveholders than to non-slaveholders (see Appendix Figure C.1), which further validates the strong signal of Confederate affinity in such name choices. In contrast, these names gained less traction among whites without Southern heritage (light red), consistent with more pervasive opposition to the Confederacy outside the South. Confederate leader names universally declined after the war, only to rise again in the early 20th century, this time among Southern and non-Southern whites alike. Strikingly, this resurgence among non-Southerners holds for those without second-generation heritage, i.e., those whose parents were not born in the South (see Appendix Figure D.1). This points to cultural spillovers beyond the diaspora, a possibility we explore later in the paper. 3 The South Goes West: Migration and Occupational Choices In this section, we use complete-count Census records and linked samples to characterize migrants’ selection, sorting, and occupational choices in the late 19th century. These results serve two purposes. First, they corroborate the historical background in Section 2. Second, they inform our causal identification strategy developed in Section 4. 12 13 The majority of these monuments, especially outside the South, were built in the early 1900s. We define prominent Confederate leaders as those linked to multiple monuments by the Southern Poverty Law Center. 9 3.1 Outmigration Using the intercensal linked records, we track migrants from each Southern origin county o to any incorporated county in the conterminous non-Southern states. For each origin county o, we total the number of white out-migrants in a given Census period through τ ∈ {1880, 1900} (1890 records were lost in a fire).14 We then estimate the following equation to characterize county-level push factors: Southern white migrantsoτ = α + push0o,τ −1 βτ + φτ populationo,τ −1 + εoτ , (1) where pusho,τ −1 is a vector of predetermined economic factors, related to industrial and agricultural activity, and ideological factors, related to slavery and the Civil War. The former include measures of manufacturing wages and output from the Census as well as cotton, tobacco, and overall agricultural potential from the Global Agro-Ecological Zones (GAEZ) database. The latter include the enslaved population share, slaveholding population share, Confederate Army enlistment rate, Civil War battle locations from Arnold (2015), and the vote share for John C. Breckinridge, the pro-slavery Southern Democratic candidate for president in 1860 (see Appendix Table A.2 for summary statistics).15 Note that the τ index on the parameters allows the push factors to differ across periods. Appendix Table A.1 highlights several economic and ideological forces driving postbellum migration. Counties with stronger support for slavery before the war—proxied by the Breckinridge vote share in 1860—exhibit less outmigration afterwards. Urban manufacturing centers also saw greater outmigration, consistent with positive individual-level selection (as shown, for instance, in Appendix Table C.3). Yet, outmigration was often triggered by economic devastation following the war, compounded by its destructive battles and the collapse of Southern agriculture. Many migrants came from agriculturally-rich areas of the Carolinas and Virginia, which were highly populated and had been major origins of poorer white migrants even prior to the war (Dippel, 2005). After emancipation, labor-intensive plantations saw their workforces freed (Prince, 2000), but many formerly enslaved people stayed on at relatively low wages, which allowed many white-owned farms to remain viable. This may explain why counties with more slaveholder families and larger Black populations saw less white outmigration. Hardship nonetheless pushed out those whites with less viable economic options but who could afford to leave. Outmigration among Former Slaveholders. We also estimate an analogous equation for former slaveholding migrants.16 Appendix Table B.1 suggests there were distinctive push factors among those 14 Concretely, we use the linked Census records together with the complete-count Census to estimate:  D  X # white men in o in τ −1 linked to d in τ Southern white migrantso,τ = × Southern whiteso,τ−1 , # white men in o in τ −1 linked to Census τ d=1 where o indicates Southern origin counties, d indicates non-Southern destination counties, and Southern whiteso,τ −1 , is based on the complete-count Census in the previous period. This allows us to approximate, for each Census period, total Southern white outmigration from o to all non-Southern counties, which we then put on the left-hand-side of equation (1). See Appendix Figure B.1 for validating evidence that the approximation works well. 15 We link Confederate Army rosters from the National Archive (via Hall et al., 2019) to the 1860 U.S. Census. 16 We use the linked Census records together with the 1860 Slave Schedule and the complete-count Census to estimate the number of former slaveholder migrants from origin county o and Census period τ :    D  X # slaveholders in o in τ −1 linked to d in τ # slaveholders in o in τ −1 linked to 1860 Census · · Southern whiteso,τ−1 , # slaveholders in o in τ −1 linked to τ # white men in o in τ −1 linked to 1860 Census d=1 where d indicates non-Southern destination counties, and Southern whiteso,τ −1 is based on the complete-count Census. 10 with ties to slavery. While several factors were common predictors of outmigration across all Southern white migrants (e.g., fewer Blacks, fewer Confederate Army households during the war, and tobacco abundance as characteristic of the Atlantic Southeast), others were distinct. Most notably, cotton abundance becomes negatively associated with former slaveholder outmigration once the cotton sector rebounds after the 1870s. Given its intensive use of slaves, cotton production severely contracted after emancipation, only to fully recover and then expand by the 1880s (Ager et al., 2021; Aldrich, 1973). 3.2 Sorting: Locations and Occupations Migrant sorting reflects a combination of economic and ideological pull factors. Here, we characterize migrants’ destinations and their occupational choices once settled. Destination Characteristics. We measure migration flows from all origin counties to all destination counties for each Census period τ . We compare sorting behavior between Southern-origin and nonSouthern-origin migrants, limiting destination counties to those outside the South and restricting the sample to individuals that moved across states. For each county pair, we use linked Census records to define an indicator for whether there were any white migrants from origin county o in τ − 1 and in destination county d in τ . We then estimate the following to describe the extensive margin of migration: P(migrantsordτ > 0) = pull0od,τ −1 β + αoτ + κrτ + εordτ , (2) where pull0od,τ −1 is a vector of predetermined destination pull factors. In addition to the economic and ideological measures from equation (1), this vector also includes geographic distance from the origin county, the preexisting share of migrants from origin county o’s state in the South, and Union Army enlistment and mortality rates in the Civil War from Dupraz and Ferrara (2021) (see Appendix Table D.1 for summary statistics). We include fixed effects for Census period separately interacted with origin county o and destination Census Division r, αoτ and κrτ , respectively. Standard errors are two-way clustered by origin and destination county. The estimates in Appendix Table A.3 again highlight the importance of economic and ideological pull factors. Certain drivers of migrant sorting were common, regardless of migrant origin. For instance, destination counties tended to be more urban, less agricultural, and relatively closer to the home county. Movers also exhibit chain migration, going to counties with relatively more previous migrants from their home states. In other ways, Southern and non-Southern migrants chose distinct locations. Southerners tended to settle places relatively early in the process of community formation. Disproportionately western, Southern white destinations tended to be more geographically distant from their origins, and density was less of a pull factor than for non-Southerners. The nascent but familiar extractive industry of the West proved attractive to many whites seeking new opportunities outside the South. Southern whites were also more likely to choose destinations sympathetic to the culture and institutions of the South (see, also, Eli et al., 2018, on postbellum ideological sorting among veterans from Kentucky). While Southern migrants gravitated towards counties with greater support for Breckinridge in 1860, non-Southern migrants generally avoided counties with legal slavery before the war.17 17 Even prior to the Civil War, “Western free states attracted non-slaveholders and pioneer families who preferred to live where slavery was not present . . . with nearly as many natives of New York living [by 1860] in Iowa as persons born in all of the slave states east of the Mississippi combined” (Dippel, 2005, p. 220). 11 Occupational Sorting into Positions of Authority. Migrants in the Confederate diaspora not only chose distinctive destinations but also entered distinctive occupations once settled. We use the completecount U.S. Census of Population to construct a dataset comprising all working-age white males living outside the South in 1900. At the time, many destination counties were still in the early stages of development, which may have created unique opportunities for entering occupations and institutions with public-facing authority, including lawyers and judges, law enforcement, public administrators, religious workers, and educators. Although few in number, these positions hold significant sway over public life. We show here that Southern white migrants strongly sorted into such occupations. Table 1 tests for differential sorting by comparing occupational choices across white male workers within the same non-Southern county. Panel (a) compares Southern migrants to all working-age, nonSouthern-born men including non-migrant natives, and panel (b) restricts the comparison group to other out-of-state migrants from non-Southern origins. The latter helps disentangle a general occupational sorting tendency among migrants from one specific to Southern migrants. Panels (c) and (d) then restrict the analysis to counties in the West and non-West Census regions, respectively, with the goal of distinguishing sorting in places with more versus less well-established institutions (see Section 5.1). Several results in Table 1 point to an outsized presence of the Confederate diaspora in local authority positions. On average, across panels (a)–(d), Southern white migrants are 35–48% more likely to work in such positions (column 1). Moreover, such sorting holds conditional on the income category of the individual’s chosen occupation (column 2). Among occupations with a similar occupational income score (occscore, a proxy for historical income), Southern white migrants are overrepresented in ones with public authority; for example, they are more likely to be religious officials than weavers or machinists. This suggests that sorting into positions of authority is unlikely to be driven solely by earnings potential or broader elite status (for additional evidence, see Appendix Figure C.2 and discussion thereof). Looking at specific occupations, we see first that migrants sort into the legal profession (column 3): they are 68% more likely to work as lawyer or judge than non-Southern-born (panel a) and 118% more likely than other migrants from non-Southern origins (panel b). This pattern holds across Western and non-Western counties. Second, Southern whites are 28% more likely to work in law enforcement compared to other migrants (column 4, panel b) and nearly 62% more likely than all other men in Western counties (panel c). Third, Southern white migrants exhibit strong sorting into religious occupations, with implied differentials ranging from 52–123% across panels (column 6). Like those in the justice system, jobs in the religious sector—and especially the clergy—often entail considerable scope for shaping public attitudes and social norms. We see more limited evidence of sorting into public administration (column 5) or education (column 7) especially in non-Western counties, perhaps due to barriers faced by Southerners residing in former Union territory. These patterns hold not only for first-generation Southern white migrants but also for their children born outside of the South (see Section 5.4). Together, the results suggest that Southern white migrants may have had a comparative advantage in or taste for authority positions. Lawyers, judges, law enforcement, and religious officials can exert considerable influence in a community, particularly at early stages of development when culture and institutions are most malleable. This may have been attractive to some whites leaving the South, drawn to opportunities for rebuilding the type of society lost during the war and threatened by Reconstruction era reforms thereafter. This may have been especially true for slaveholding elites, who had particularly strong affinity for the Confederacy (recall Appendix Figure C.1). 12 Sorting Among Former Slaveholder Migrants. For slaveholders, ideology and economics were deeply entwined; they had benefitted the most from slavery and thus lost the most from emancipation. The West presented opportunities for former slaveholders to preserve the antebellum hierarchies of the South in a new region rife with extractive potential. Many of these migrants likely fit the stereotype of the aggrieved Southern white “who hated that racial equality could be enforced by the government [and] saw the West as the only free place left in America” (Richardson, 2020, p. 9). We show here that indeed, relative to the broader Confederate diaspora, former slaveholders chose a distinctive set of destinations and occupations consistent with their quest to gain hold of these malleable spaces. Our analysis relies on the linked sample described in Section 2.1, and we restrict attention to men living in the South in 1860 and outside the South in the decades thereafter. Former slaveholders comprise just over 6% of this linked sample of Confederate migrants. We compare slaveholders’ choices to those of non-slaveholders from the same origin county in the South. We find that slaveholders gravitated towards Western counties that looked economically and politically similar to the South. The coefficients in Appendix Table C.1 identify the differential likelihood of choosing a particular type of destination county across slaveholder and non-slaveholder migrants leaving the same origin county in the same Census period. The counties chosen by former slaveholders were 20% more likely to be in the West (column 1), had 11% lower population density (column 2), were 16% more likely to be cotton-suitable (column 3), had 7% higher Breckinridge vote shares in 1860 (column 4), had 6% fewer Union Army soldiers per capita during the war (column 5), and nearly 42% more slaves on average just prior to the war (column 6). These patterns hold, in panel (a), for the original slaveholder (typically the household head) and, in panel (b), for a broader group that includes other white men living as of 1860 in slaveholding households (typically the children of slaveholders). Former enslavers also differentially sorted into positions of authority outside the South. Appendix Table C.2 shows that, compared to the average Southern white migrant from the same origin county, former slaveholders were about 50% more likely to work in any occupation of public-facing authority (column 1). As with the general diaspora, such sorting does not appear to be solely driven by earnings opportunities in these positions of power (column 2; also see Appendix Figure C.2). Some of these former slaveholders may simply be continuing the same line of work they had in the South prior to the war: Appendix Table C.3 shows that among former slaveholders, those who left the South were more likely to have worked in positions of authority prior to the war. Breaking out specific occupations, former slaveholder migrants were more than twice as likely to work as lawyers or judges (column 3) and nearly three times as likely to work in public administration (column 5) as the average non-slaveholding Southern white migrant. They were no more likely to work in the law enforcement, religious, or education sectors (columns 4, 6, and 7, respectively). All patterns are similar for slaveholders alone (panel a) and inclusive of their male kin (panel b). In sum, we find overrepresentation of former slaveholders in underdeveloped counties and in positions of authority outside the South. Many of these erstwhile Southern elites may have played a leadership role in the Confederate diaspora. When one accounts for the small number of these authority positions, combined with the strong occupational sorting by former slaveholders therein, it becomes clear that former enslavers and their kin could play a large role in local institutions in many places outside the South (see Appendix Figure C.3 for maps with back-of-the-envelope estimates of occupational 13 shares by county).18 In the following section, we explore the early impacts of the entire Confederate diaspora and then, in Section 5, explore the distinct impacts of former slaveholders. The sorting patterns described above will later help clarify how a small number of postbellum migrants from the South could shape the geography of cultural change across America by the early 20th century. 4 Diffusion of Confederate Memory and Racial Norms This section explores how white migrants from the U.S. South after the Civil War helped spread Confederate memory and norms across the westward-expanding country. Our results suggest that this diaspora laid the groundwork for a resurgent Confederate culture to gain traction across America in the early 20th century. We show that besides their role in the national expansion of Southern-born organizations like the UDC and the KKK (and associated expressions of racial animus), these migrants also helped boost a novel form of racial exclusion that was not as prevalent in the South—the sundown town. 4.1 Identification Strategy We identify county-level effects of the Confederate diaspora using a shift-share instrumental variable (SSIV) framework developed in this section. Our primary second-stage estimating equation is: yc = αs + β · % Southern Whitesc,1900 + X0c γ + εc , (3) where the outcomes yc span four measures of Confederate culture in non-Southern county c in the early 1900s: any Confederate memorials, UDC chapters, 2nd KKK chapters, and lynchings of Blacks. We also consider a composite Confederate Culture Index (CCI), which sums these indicators. Figure 3 shows CCI scores (from 0 to 4) by county, and Appendix Figure D.2 shows maps for each outcome along with available time variation for the KKK and lynchings, which we use in secondary analyses. Our primary regressor, % Southern W hitesc,1900 , measures the Southern-born white population share in county c in 1900 observed in the complete-count Census. This variable captures the postbellum Confederate diaspora, while predating the larger white outflows during the Great Migration after 1900.19 In some specifications, we also include a distinct measure for the population share of Southern-born former slaveholders in county c in 1900 based on various linked Censuses from 1860–1900. We cluster standard errors across counties within 60 × 60 mile grid cells along the lines of Bester et al. (2011) and show robustness to other forms of spatial autocorrelation (Adao et al., 2019; Conley, 1999). There are two interrelated threats to causal identification of β in equation (3). First, place-specific factors, such as factor endowments conducive to plantation labor, may confound interpretation. Second, endogenous location choices based on previous settlement patterns may bias OLS estimates. Depending on the relative importance of ideological and economic sorting—both of which were salient, as we showed in Section 3—the bias could go in either direction. Economic sorting would imply downward bias of OLS estimates, insofar as economically-vibrant locales would have attracted a broad array of 18 For example, our estimates suggest that, as of 1900, former enslavers and their kin comprise 7.5% of governance occupations and 8.1% of religious occupations in the Pacific and Mountain West Census divisions compared to 1.3% and 0.8% in the New England and Middle Atlantic divisions. 19 This time horizon focuses our analysis on the group of interest: those who experienced slavery, the Confederacy, and its loss and then moved in the years after the war. In our baseline, we consider migrants from the former Confederate states and territories to the conterminous non-Southern U.S.; we then vary the set of origin and destination states for robustness. 14 migrants beyond Southerners with an affinity for Confederate culture. Ideological sorting, on the other hand, would have favored culturally-proximate destinations, implying an upward bias. We address these concerns in three ways. First, equation (3) includes state fixed effects, αs , and many controls, Xc , to absorb confounders. Our baseline includes log population in 1870, as well as log county area in contemporary boundaries. We also include additional sorting correlates, motivated by the relevant pull and push factors discussed above: cotton, tobacco, and overall agricultural potential, Black population shares in 1870, enslaved population shares in 1860, Breckinridge vote share in 1860, Union Army enlistment and mortality rates in the Civil War, and indicators for frontier status in 1860 and from 1860–89. Continuous controls are entered flexibly, using quadratic terms. Further robustness checks adopt a Belloni et al. (2014) double LASSO procedure to select optimal controls among this set. Second, we adopt a shift-share IV strategy that combines two sources of variation: shares and shifts. The shares are based on the cross-sectional distribution of white migrants from Southern origin state j living in non-Southern county c in 1870, which we denote πjc,1870 . The shifts are based on the change in the number of whites from Southern state j living outside the South between 1870 and 1900, which we denote ∆Mj,1870−1900 . Following Bazzi et al. (forthcoming) and building on Boustan (2010) and d j,1870−1900 , based on origin-county push factors over Derenoncourt (2022), we use predicted shifts, ∆M the 1870–1900 period. Together, these predict the stock of Southern white migrants in 1900 as: Zc,1900 = J X d j,1870−1900 . πjc,1870 ∆M (4) j=1 Scaling Zc,1900 by the 1870 county population yields our SSIV for % Southern W hitesc,1900 in equation (3). This IV isolates the contribution of the Confederate diaspora that is due to changes in Southern white inflows during the postbellum period from 1870 to 1900. The SSIV is relevant to the extent that postbellum whites tended to follow the migratory pathways introduced by their antebellum predecessors, as discussed in the previous section. The share component, πjc,1870 , reflects these historical networks at the dawn of the postbellum period for all counties c incorporated in the U.S. by 1870. Because the shares alone may be endogenous, our SSIV combines them with predicted shifts, based on origin-county push factors. This “push factor” version of the standard SSIV can satisfy the exclusion restriction even when the share component is endogenous, to the extent that the shift is based on exoged j,1870−1900 , by summing nous shocks (see Borusyak et al., 2022). We construct our predicted shift, ∆M V predicted Southern white outmigration, Southern white migrantsoτ , for each origin county o for Census periods 1870–1880 and 1880–1900 based on equation (1). Specifically, we use a flexible LASSO algorithm, which shrinks the set of origin-county predictors as well as their square and cross-term interactions into an optimal subset, to predict Southern outflows for each origin-county-period. These are d j,1870−1900 in equation (4): then aggregated to the sending state j level to produce ∆M X X 1900 V d j,1870−1900 = ∆M Southern white migrantsoτ , (5) o∈j τ =1880 Later, we use a variant of equation (1) to predict a distinct set of shifts for Southern former slaveholders. Third, we deploy several additional checks that address residual concerns. We include in Xc the Southern white population share in 1870 to rule out a confounding influence of antebellum migrants. 15 Controlling for % Southern W hitesc,1870 in equation (3) renders the IV specification equivalent to one with the change in Southern white shares between 1870 and 1900 as the regressor of interest, thus absorbing time-invariant heterogeneity. Later, alongside this control, we also implement the Adao et al. (2019) random-shifts placebo exercise to clarify the identifying variation in the SSIV. Finally, we re-estimate equation (3) on a restricted sample of counties with < 2 people per square mile in the 1860 Census. Such places had more limited scope for antebellum populations to influence culture and institutions independent of the postbellum Confederate diaspora. Together, these checks help ensure that early migrant shares are not confounding our causal interpretation. 4.2 Confederate Diaspora Impacts in the Early 20th Century We now report our main county-level findings on the cultural influence of the Confederate diaspora. We begin by identifying the impacts of these early postbellum migrants on the reinvigoration of Confederate narratives in public life and then turn to more overt forms of racial animus and white supremacy. Confederate Memorialization and Lost Cause Advocacy. After the Civil War, memorialization efforts spread throughout the former Confederacy and, before long, into the former border states and large swathes of the “Old West,” too. Table 2 shows that white migrants from the postbellum South hastened the diffusion of Confederate nostalgia throughout the country. We examine two indicators for any (i) Confederate memorials (i.e., monuments, location names), and (ii) UDC chapters after 1900.20 For memorials, OLS estimates, with state FE and flexible controls for county population and area, suggest that a 1 p.p. increase in the share of Southern whites in 1900 (relative to a mean of 2.2%) is associated with a 0.8 p.p. increase in the likelihood of Confederate memorialization (relative to a mean of 25% outside the South). The corresponding IV estimate is over twice as large (column 2) and is robust to controlling for the 1870 share of Southern whites (column 3). Estimates for UDC chapters, in columns 4–6, are similar but imply larger effect sizes given the relatively limited presence of the organization outside the South (only 10% of counties). The estimates in Table 2 corroborate the insights of historians who note an “outsized cultural influence” of migrants from the South, particularly in the West, because of organizations like the UDC (Waite, 2020, p. 34). All estimates are significant at conventional levels, with evidence of a strong first stage and robustness to potential instrument weakness. Expressions of White Supremacy. The late 19th and early 20th centuries also saw the spread of more overt expressions of racial animus that had historically been associated with the South. Most prominently, the KKK reemerged in 1915, following The Birth of a Nation’s commercial success in propagating Lost Cause narratives and a rosy image of the original KKK. Although the 2nd KKK had been founded in Georgia, the organization quickly spread throughout the West and Midwest during the 1920s (McVeigh, 2009). Lynchings of Black people also spread after the war (Blocker et al., 2013). Such killings often took on a symbolic role: to the extent that they were perpetrated publicly, lynchings signaled local commitment to white supremacy (Henderson et al., 2021). 20 We identify likely Confederate location names (i.e., place, street, school) using the more restrictive set of “distinguishing names” as in panel (b) of Figure 2, so as to minimize the potential for false positives in our county-level analysis. In Appendix Table B.2, we go even further by restricting to those names with at least two (columns 1–3) and at least three (columns 4–6) words. These reduce outcome variation but produce similar estimates. In Appendix Table B.3, we show that the IV estimates hold across distinct memorial types in (i) monuments, (ii) place names, and (iii) street names. 16 Table 3 shows that the postbellum Confederate diaspora played an important role in helping to spread such expressions of racial animus beyond the South. We examine two binary outcomes for any (i) KKK chapters, and (ii) lynchings of Black people after 1900. For KKK chapters, which could be found in 37% of non-Southern counties from 1915–40, OLS estimates suggest that a 1 p.p. increase in the share of Southern whites in 1900 is associated with a 0.8 p.p. increase in the likelihood of a KKK chapter (column 1). The corresponding IV estimate is roughly twice as large (column 2) and is also robust to controlling for the 1870 share of Southern whites (column 3). Coefficients for lynchings are similar yet even more striking, given the rarity of lynchings outside the South after 1900 (5% of counties). As before, IV estimates (1.5 p.p., column 5) are larger than OLS (0.8 p.p., column 4) and robust to controlling for earlier migration in the 1870 shares (column 6). This may be due, in part, to the greater cultural attachments, or deeper racial grievances, of those who directly experienced Confederate secession and defeat. Moreover, these results are unique to Black lynchings as we find no diaspora effect on white lynchings (see Appendix Table B.4). This helps distinguish racially-motivated attacks from a generalized culture of violence associated with the South (Grosjean, 2014; Tan, 2023). Together, the findings in Tables 2 and 3 shed new light on the co-location of Confederate memorialization and more overt and violent forms of racial animus. Our findings complement Ang (2023), who links exposure to The Birth of a Nation showings during its early roadshow to increased lynchings and KKK activity. Our findings together suggest a close relationship between Confederate iconography in its advent and the entrenchment of white supremacy in communities across America. Interpreting Magnitudes. The estimates in Tables 2 and 3 suggest a large diaspora impact consistent with direct influence over important public-facing institutions as well as cultural transmission to nonSouthern populations. We explore these channels of influence in Section 5, and below we provide further robustness checks towards a causal interpretation. Before proceeding, however, it is worth dwelling briefly on why the IV is consistently larger than the OLS, even though they are not always statistically significantly different. Although noise in historical data attenuates OLS estimates, it is unlikely to fully explain the differences. Two other factors seem plausible. First, economic sorting may have been relevant. To the extent that Southern whites chose more productive locales, which were also likely to have attracted a culturally-diverse array of migrants besides Southern whites, the impact of Southern white migrants would be diluted and OLS estimates biased downward. Indeed, Section 3 points to significant economic sorting, but also to ideological sorting, which would imply upward bias. Our SSIV addresses both types of endogeneity. Moreover, our SSIV identifies a particular local average treatment effect (LATE): counties with the strongest chain migration from the South, and hence those underpinning the SSIV, are perhaps most likely to have migrants attached to Confederate culture. Such a pattern is consistent with other recent SSIV applications (Calderon et al., 2019; Derenoncourt, 2022). Following Goldsmith-Pinkham et al. (2020), a LATE interpretation for SSIVs rests on positive “Rotemberg weights,” which capture the relative contribution of different origin states to the second-stage identifying variation. In our case, the vast majority (97%) are positive. To further ensure that no state is driving our effects nor undermining a LATE-based interpretation, Appendix Figure B.3 drops sending and receiving states one-by-one, using the composite Confederate cultural index (CCI) as the outcome. While effect sizes vary in sensible ways (e.g., California is an important destination), all remain significant at conventional levels. 17 Additional Checks. This section bolsters the case for causal identification. We summarize several key checks and report most in Table 4, for the four core outcomes and the composite CCI. Alternative Standard Errors. First, we show that inference is robust to alternative forms of spatial correlation, across nearby counties (Conley, 1999) or counties with similar initial shares in the SSIV (Adao et al., 2019). These deliver similar takeaways as our baseline (see row 1 of Table 4). Varying Controls. Subsequent rows of Table 4 examine robustness to alternative controls, Xc , in equation (3). Estimates remain sizable, statistically significant, and stable across the following specifications: controlling for 1870 Southern white shares only (row 2); controlling for sorting correlates, including cotton, tobacco, and overall agricultural potential, Black shares in 1870, slave shares in 1860, Breckinridge vote in 1860, Union Army enlistment and mortality, and frontier status (row 3); and using a double LASSO control selection on rows 1 and 3 (rows 4 and 5, respectively). We further clarify robustness to differential spatial structures, including no state FE (row 6) and reweighting by population size (row 7). Varying SSIV Shifts. Although we use LASSO to select the push factors that predict the shifts in our SSIV, the algorithm ultimately selects from those that we initially provide. To ensure that effects are d j,1870−1900 , with the not an artifact of our choice of push factors, we replace our predicted shifts, ∆M actual shifts, ∆M j,1870−1900 , in equation (4). Again, estimates are relatively unchanged (row 8). We also implement the Adao et al. (2019) diagnostic, in which we replace predicted shifts with randomlygenerated ones. Doing so yields < 5% of coefficients statistically significant at that level.21 This ensures that the (potentially endogenous) shares are not the sole identifying variation in our SSIV. Early Migrants and Critical Junctures. We further address concerns about confounding effects of early, antebellum Southern white migrants, as well as exclusion restriction violations via their influence on initial culture and institutions. Here, we focus on a restricted sample of relatively “unsettled” counties, with < 2 residents in the 1860 U.S. Census. These estimates, in row 9 of Table 4, are somewhat larger than baseline ones, which suggests that sorting towards counties on the frontier—at a critical early juncture in their histories—may have helped solidify the cultural influence of postbellum diaspora whites. Below, we provide further evidence supporting this interpretation. (Re-)Defining Sending and Receiving States. Our findings suggest that early postbellum migrants brought a distinctive set of Confederate cultural attachments to their destinations outside the South. It is thus important to ask whether our results are contingent on the particular demarcation of the South. We want to clarify, for example, whether results depend on cultural complementarities with residents of border states, or if those border states themselves had deep antebellum Confederate cultural proclivities. Importantly, our core estimates are not particularly sensitive to our choice of origin states. Among the 12, only Oklahoma was not part of the Confederacy; we include it in our baseline measure, as the state (and namely the Indian Territories before it) had maintained legal slavery and formal treaties of alliance with the Confederacy during the Civil War.22 Given Oklahoma’s unique history, row 10 of Table 4 excludes it from the set of sending states with little change to our estimates. Besides Oklahoma, Missouri and Kentucky also had close ties to the Confederacy; beyond a shared history of slavery, both states had shadow governments declaring their allegiances to the Confederate Army during the war. In 21 In equation (4), we interact the origin-state shares from 1870 with shifts drawn from a random normal distribution of mean 0 and variance 5. We then re-estimate the baseline 1,000 times. See Appendix Figure B.4 for a coefficient plot. 22 Figure 3 illustrates the cultural similarities of Oklahoma to the Confederate states in the postbellum period. 18 row 11, we include Missouri and Kentucky among sending states rather than destination states. Using this re-defined set of Southern white migrants decreases estimates somewhat, consistent with weaker affinity for Confederate culture among migrants from non-Confederate-controlled states. To ensure that our findings are not simply driven by places with antebellum slavery outside the South, row 12 of Table 4 excludes the five border states of Delaware, Kentucky, Maryland, Missouri, and West Virginia. Slavery was legal in these states before the war, and they were popular destinations for Confederate migrants thereafter (see Figure 1). Although this restriction reduces the variation in some outcomes, particularly lynchings, our estimates uniformly increase, with large diaspora effects even in places with no ties to slavery and with significant cultural distance from the Confederacy. Northern Migration. The West was settled not only by white migrants not only from the South but also from former Union states. This raises the question as to whether Confederate memorialization and antiBlack sentiment were driven by white migrants in general or whether Southern whites had a distinctive and outsized impact. Table 5 provides evidence consistent with the latter, focusing on Western states to which migrants from former Union and Confederate territories flocked after the war. KKK activity is the only outcome from Tables 2 and 3 that responded significantly to migration from the North. As noted earlier, some of the northern affinity for the Klan concentrated around negative sentiment towards Catholic Irish and Italian immigrants, as well as Jews (Fouka et al., 2022; McVeigh, 2009). Consistent with this history, we find little impact of migrants from former Union states on Confederate memorialization or on lynchings of Black people. 4.3 A New Form of Racial Exclusion: Diaspora Impacts on Sundown Towns While the diaspora effectively transmitted Confederate norms, they also helped to build new racial norms without roots in the South. This section shows how postbellum migrants contributed to novel modes of racial exclusion in their destination communities, in the form of “sundown towns.” Distinct from the de jure exclusionary institutions associated with the antebellum South and later Jim Crow era, sundown towns emerged and proliferated mainly outside the South. Named after bans on Black and other minority populations’ presence within town limits after sunset, this informal institution diffused widely from 1890 to 1960 (Crowe, 2012; Loewen, 2005; O’Connell, 2019). Prior to the Civil War, states like Ohio and Oregon tried to preclude Blacks from settling there, but the federal government blocked such large-scale exclusion efforts. Sundown towns could be seen as a postbellum innovation that facilitated exclusion at a scale small enough to evade federal attention or control. Although concentrated outside of the South, sundown towns were more likely to form in places where large numbers of Southern white migrants settled. Diaspora whites often perceived Black people as potential sources of economic and political competition where they lived, prompting new forms of de facto exclusion outside of the South (Cook et al., 2022; Waite, 2021). At the same time, sundown towns often went hand-in-hand with Klan and lynching activity, as we show in Bazzi et al. (2022) and consistent with graphic accounts of incipient sundown towns in Loewen (2005). These activities facilitated exclusion of Blacks on a town-wide basis by both Southern and non-Southern whites. Table 6 explores the influence of the Confederate diaspora on sundown towns using two measures.23 The first relies on a measure from Loewen (2005) (via Taylor, 2020), based on the centroids of documented sundown towns, aggregated to the county level. The second is based on town-level data from 23 These results build upon and extend initial findings in our AEA Papers & Proceedings article (Bazzi et al., 2022). 19 the Census Place Project (Berkes et al., 2022), with which we build a more granular measure of what Loewen calls the “Great Retreat” of Blacks from towns and counties across America (see Figure D.3 on the pervasive Black depopulation of towns in the early 1900s). Together, these two measures provide a rich and more complete picture of Black exclusion and depopulation across time and space. The estimates in Table 6 point to a sizable diaspora imprint on the geography of sundown towns. SSIV estimates suggest that a 1 p.p. increase in the diaspora share in 1900 is associated with a 22% increase in the number of sundown towns in the county (column 1). We find similar results using our second, town-level regression. Relative to the sample mean, a 1 p.p. increase in the town-level diaspora in 1900 leads to a 2.6 p.p. increase in the likelihood that a non-Southern town with ≥ 25 Black residents as of 1870 had zero Black residents after 1900 (column 4). Effects are smaller and insignificant among non-Southern towns that already had few Black people as of 1870. These findings are robust to using county-level measures of Southern white migrant shares (column 5–6). And while some of the depopulation episodes in this town-level sample may stem from voluntary exodus, rather than forced exclusion, it is reassuring that our town-level results mirror column 2, which adapts the town-level specification to the county-level Loewen (2005) sample of confirmed sundown towns. In sum, many whites leaving the postbellum South chose destinations with few Black people, while others helped build new means of removing and excluding non-white populations. This contributed to the widespread “Great Retreat” of Blacks during the early-to-mid 20th century, while also ensuring little minority opposition to white supremacy in the decades to come. 5 Channels: Institutions, Organizations, and Culture The early 20th century ushered in a wave of national reconciliation as organizations like the KKK bridged the North–South divide and Lost Cause ideology gained traction in new arenas beyond the South. This section digs deeper into this process to understand, at a micro level, how the Confederate diaspora changed the trajectory of postbellum nation building. First, we show that migrants more effectively diffused Confederate culture in places lacking established culture and institutions on the eve of the Civil War. Second, we find that, within the diaspora, those with personal ties to slavery played an outsized role in propagating Confederate memory and entrenching white supremacy in nascent institutions. Third, we use individual-level data to show that first-generation migrants from the South transmitted Confederate culture not only to their children born outside the South but also to their non-Southern neighbors. Fourth, we show that the second-generation followed in their parents’ footsteps within the halls of power, maintaining the diaspora’s overrepresentation in positions of authority and even entering new domains like education where their parents had not gained a foothold. Together, these results further clarify how white migrants from the postbellum South not only transplanted Confederate norms but also captured early institutions ensuring the perpetuation of those norms over the long run. 5.1 Early Settlement in Malleable Places The influence of the Confederate diaspora may have varied across destinations depending on the support or opposition of incumbent residents and local institutions. It could have been easier for a small group of Southern migrants to stoke cultural change in counties without well-developed institutions, with small populations, or with residents favorably predisposed to Confederate ideology. By contrast, 20 these migrants could have faced greater resistance in counties with deeply ingrained norms and strong institutions, especially when those norms and institutions were built by those with antipathy towards to the former Confederacy. Table 7 provides evidence along these lines by exploring two key sources of place-based heterogeneity in the impact of the diaspora. In each case, we augment equation (3) to include the given heterogeneity measure, call it hetc , and its interaction with % Southern W hitesc,1900 , instrumenting that interaction with SSIV×hetc .24 Panel (a) shows that the Confederate diaspora exerted greater influence in frontier communities where both culture and institutions had not yet taken deep root on the eve of the Civil War. Concretely, we see that the postbellum diaspora was more effective at transmitting Confederate memory and building white supremacist organizations in counties that were still on the frontier in 1860. The estimate for the composite CCI outcome in column 5 suggests that such efforts were three times more effective on the frontier than in more densely settled counties where the diaspora imprint nevertheless also materialized (as seen from the estimates in row 1). This is consistent with the Confederate diaspora having greater scope for ideological transmission in places without well-established institutions or culturally-dominant groups. We find very similar estimates using another proxy for undeveloped social infrastructure: an indicator for whether the county was not yet incorporated into the U.S. by 1850.25 Panel (b) shows that the Confederate diaspora exerted weaker influence in communities where a greater share of residents enlisted in the Union Army during the prior years of the Civil War. Many in these communities would have lost family members in the fight against the Confederate Army. It is not surprising, then, that the Confederate diaspora was less effective at transmitting its norms in these places. The summary estimate for the CCI in column 5 suggests a nearly exact countervailing effect of Union enlistment: going from zero to 100% Union Army enlistment rate wholly nullifies the impact of the Confederate diaspora. At the same time, the estimates in row 1 suggest that the Confederate diaspora had its largest impact in communities with zero participation in the Union Army, some of which would have been in frontier areas of the country. Together, the results in Table 7 provide empirical support for an overarching narrative in the paper: the Confederate diaspora was most effective at shaping norms in areas of the country still early in the process of nation- and state-building or that lacked strong oppositional (i.e., pro-Union) cultural forces. It was places like these where key social institutions could be contested by early settlers and where a diaspora with strong cultural entrepreneurs might have unique pathways to public influence. Postbellum migrants from the South indeed gravitated towards some of these places (see Section 3.2), and the estimates in Table 7 suggest that their diaspora communities causally transmitted a strong Confederate legacy just as these areas became part of a now-larger United States. 5.2 Former Slaveholders and the Foundations of Local Institutions Just as the diaspora was more influential in certain places, so too were certain elite members of the diaspora across destinations. Of the nearly one million white migrants who left the South in the three decades following the Civil War, only about 6.7% had personally held slaves in 1860, with another 13.9% coming from slaveholding households. In this section, we show that this minority former slaveholding elite in the Confederate diaspora had a distinctive influence on culture and institutions at destination. We 24 25 Both measures of heterogeneity are included among the additional controls used in row 3 of Table 4. By 1850, 604 counties outside the South were not yet incorporated into the U.S., and 553 of these were on the frontier. 21 develop a variant of the SSIV for former slaveholder migrants, using shifts based on former slaveholder migration from the South and distinct push factor variation over Census periods from 1870 through 1900. We then re-run key regressions from Tables 2 and 3, as well as regressions with the composite CCI outcome, in each case distinguishing former slaveholders from the general diaspora. In columns 1–2 of Table 8, we regress outcomes separately on the total shares in 1900 of Southern white migrants (mean of 2.2%) and linked former slaveholder migrants (mean of 0.3%), respectively. Both columns include baseline controls as well as the corresponding 1870 shares. Once scaled by their sample means, these effects are of comparable size to each other: an increase from zero to the sample mean of either is associated with about a 0.3 point increase in a non-Southern county’s CCI (relative to a sample mean of 0.76, on a scale from 0 to 4). When these regressors are included and instrumented jointly, the former slaveholder migrant effect dominates the overall migrant effect. Column 4 reports small and imprecise associations with Southern whites in general, but large and precise effects associated with former slaveholder migrants: an increase from zero to the latter’s sample mean is associated with a 0.8 point increase in the CCI. This pattern holds across all four outcomes from Tables 2 and 3, with varying precision. Importantly, there is sufficiently distinct identifying variation in the overall diaspora and slaveholder SSIVs to generate individually and jointly strong first stages (see the Sanderson and Windmeijer (2016) and Kleibergen and Paap (2006) diagnostics in Table 8). This distinct variation stems from the differences in outmigration and sorting across slaveholder and non-slaveholder migrants from the South, as described in Section 3.2. Additional columns of Table 8 address measurement and endogeneity concerns. One concern is that the slaveholder-migrant effects are confounded by antebellum slavery institutions in the places to which those migrants gravitated.26 We address this in columns 3 and 5 by controlling for the share non-Southern slaveholders in 1860. Estimates are generally robust, consistent with a distinct influence of slaveholders hailing from Confederate-controlled areas. Column 6 addresses a measurement concern: we measure postbellum Southern white migration with linked Census records, mirroring the measurement of slaveholder migration. Column 7 addresses residual concerns about place-based confounders, given the endogenous sorting patterns we document in Section 3, by including the sorting correlates outlined in Section 4. Results are robust to these alternative specifications. Overall, these findings suggest that those in the diaspora with personal ties to slavery were instrumental to the diffusion of Confederate culture. While this does not preclude the importance of the average migrant in this process, it does suggest that former slaveholders played an essential leadership role within the diaspora, ensuring its cultural imprint. Insofar as former slaveowners shaped the local institutions and racial norms that would attract subsequent inflows of other Southern whites in the years to come, this may explain why the effects mostly load onto the former as of 1900. It is also possible that the presence of former slaveowners, through their crucial effects on institutions, was a necessary condition for less powerful members of the diaspora to influence culture. The key role of former slaveholders in the diaspora resonates with historical accounts of slaveholder influence in society. Dippel (2005, p. 218), for instance, notes that although “they were a small minority, large slaveholders invariably came to control the political and economic systems” in the places they settled. This influence extended to the postbellum period, in which former slaveholder elites who left 26 In some non-Southern counties (e.g., those in border states), native former slaveholders would have outnumbered migrant ones. As an additional check, we drop border states from our receiving states in Table 4, which increases effect sizes. 22 the South explain much of the impacts of the Confederate diaspora. We turn now to understanding how such an elite imprint materialized and its implications for Blacks living outside the South. Occupational Sorting and Institutional Influence. Despite being a minority, former slaveholder migrants had a sizable influence in the postbellum diaspora, according to estimates in Table 8. This is consistent with their sorting patterns, which favored destinations in the country’s underdeveloped West, as well as public-facing positions of authority therein (see Sections 3 and 5.1). Whereas Southern white migrants generally displayed such tendencies, former slaveholder migrants displayed even starker sorting into certain governance positions, such as in public administration. These results earlier in the paper set the stage for our analysis here. Table 9 reveals a large and distinct imprint from former slaveholding migrants working in positions of power.27 We augment the specification from Table 8 with binary indicators for whether any former slaveholding migrants are working in one of the occupations considered in Table 1. Columns 1–7 use our composite Confederate cultural index (CCI, scored 0–4) as the outcome. We find that the CCI is significantly higher in counties with at least one former slaveholding migrant working in public administration or a religious occupation. When breaking out the CCI components in columns 8–11, we find patterns of heterogeneity that resonate with the historical record on institutional support for Confederate culture and racial norms. Memorials are more likely in counties with former slaveholders working in law enforcement; UDC chapters are more likely where former slaveholders are working in public administration; and KKK chapters are more likely where former slaveholders are working in public administration and especially in religious institutions, the latter being a core element of Klan life and its overt Protestant chauvinism (Goldberg, 1981). With the exception of education occupations, coefficients are positive across all public-facing authority occupations and all outcomes.28 Importantly, such impacts go above and beyond the general contribution of Southern white and former slaveholder migrants, both of which are accounted for, and instrumented, in the regression. These results are also not an artifact of occupational sorting by all diaspora whites. While we find similar patterns for any Southern whites working in these occupations, these largely disappear once we control for, and instrument, former slaveholder shares (see Appendix Tables C.4 and C.5). To further corroborate the distinctive role of former slaveholders in authority positions, we consider several placebo occupations where one would not expect much influence on public institutions or visibility in public life more generally. Appendix Table C.6 includes, alongside each authority occupation, a placebo one with similar earnings potential and overall employment shares but with no such component of power or influence. None of these placebo occupations prove to be as important, with estimates that are small and insignificant across specifications. Despite such reassurances, the findings in Table 9 should ultimately be seen as descriptive, insofar as the causal effects of occupational sorting would be difficult to convincingly identify. Yet, in conjunction with our slaveholder SSIV strategy and estimates in Table 8 above, these results strongly suggest that Southern white migrants—and former slaveholders in particular—not only imported Confederate culture directly but also shaped institutions in their destinations, thus solidifying their posthumous legacy. 27 Given the dearth of former slaveholder migrants who can be successfully linked across Census periods to specific occupational categories within specific destination counties as of 1900, for this exercise we define migrants with ties to slavery as those white men who lived in Southern slaveholder households in 1860. One caveat is that this also includes family members who may be less sympathetic to slavery even though they exhibit similar sorting patterns (see Section 3). 28 The null for education is perhaps because such occupations were dominated by women, whereas linked individuals are men. 23 Micro Effects of Former Slaveholder Governance. The results thus far suggest that occupational sorting by the Southern former slaveholding class may have helped to entrench Confederate culture throughout the country by the early 20th century. What remains to be seen is whether such influence matters for the well-being of non-white minorities. Here, we explore how former slaveholder involvement in local governance affected individual outcomes among Black residents. We explore this question by constructing a dataset of young Black and (non-Southern) white men in 1900, along with their exposure to Southern white migrants from slaveholding backgrounds working in governance (i.e., policing, legal, or administrative occupations). We then link these men to the 1920 Census, which contains information on incarceration status. We restrict the sample to men through the age of 25 in 1900, who were at peak probability of being incarcerated in the intervening period (Eriksson, 2019, 2020). We regress incarceration status on an individual race indicator (Blackic = 1), which we interact with an indicator for whether an individual’s county, as of 1900, had any former slaveholding migrants working in local governance positions. In addition to basic individual-level controls, specifications variously include fixed effects for state or county of residence in 1900, with the latter fully absorbing the overall effects of the Confederate diaspora and former slaveholders therein.29 Table 10 estimates these probabilities of incarceration, which always depend strongly on race. Black males living in 1900 in a non-Southern county without former slaveholding migrants working in governance were nearly 5 times more likely than the average white male to be incarcerated from 1900–20. That probability roughly doubles for Black men living in counties with such migrants. These suggestive estimates illustrate an important micro-level consequence of local “institutional capture” by the former slaveholding class of the diaspora. 5.3 Cultural Transmission: Families, Neighbors, and KKK Mobilization Beyond its entrenchment within local institutions, Confederate culture was also transmitted directly throughout the country by individuals and, especially, new organizations. The latter most prominently included the UDC in the 1900s–20s and the 2nd KKK in the 1910s–30s. Whereas the former was composed primarily of Confederate whites and their direct descendants (Cox, 2003), the latter had a largely non-Southern member base. The 2nd KKK was first organized in Georgia, before finding popularity throughout the Midwest and West, where “as many as six million Americans heeded its call to resist Catholics, Jews, lawbreakers, Blacks, and immigrants” at its peak in the 1920s (Goldberg, 1981, p. 8). At the same time, the group relied heavily on “rituals and symbols designed to memorialize the Confederacy, Southern-style chivalry, and white Protestant supremacy,” helping to bring those ideals and iconography to new audiences across the country (Gregory, 2005, p. 294). Table 3 provided evidence consistent with a causal effect of the diaspora on KKK chapter formation. In this section, we go deeper to show, at the individual level, how the diaspora mobilized white Americans to join this organization. We explore this mobilization process using the complete membership data from the Denver, Colorado chapter of the KKK in the 1920s.30 Colorado had been at the epicenter of the 1920s Klan, with tens of thousands of members, including the Denver city attorney and chief of police (Goldberg, 1981). 29 Individuals are often incarcerated in counties far from their county of residence-cum-arrest. This is one reason why we allow for a lag between time of exposure to Southern whites and time of observed incarceration. Another is that it may take time for the cumulative adverse effects of exposure to Confederate culture and institutions to materialize in incarceration. Appendix Table C.7 shows robustness to additional fixed effects for the individual’s contemporaneous county of residence in 1920. 30 We follow Fryer Jr. and Levitt (2012) in using these data, although we expand record matching to the complete-count Census. 24 To identify diaspora connections, we construct a dataset of all white men in the 1920 Census living in the Denver metropolitan area. We then match these men to the membership records from the mid-1920s. Using the ABE algorithm from Abramitzky et al. (2019) with NYSIIS standardized names, we find 21.3% percent of all white men—about 51,000 individuals—in these records.31 We then explore this matched-KKK-membership outcome for different white sub-populations in the Denver area. Table 11 provides a uniquely rich window into the transmission of KKK activity within and beyond the diaspora. First, in panel (a), we find that the first- and second-generation diaspora was overrepresented in the Denver KKK. We begin by focusing on Denver county where 24% of white men without Southern heritage belong to the KKK. By comparison, Southern-born men are 3.2 p.p. more likely to be KKK members, and second-generation men (i.e., those with at least one Southern-born parent) are 3.7 p.p. more likely (column 1). The latter arises for both Southern-born mothers and fathers (column 2), and all of these estimates are significant at the 1 percent level, while clustering standard errors by neighborhood or enumeration district, of which there are 313 in the county. Columns 3–6 expand the sample to include individuals in all 14 Denver metro area counties, which nearly doubles the sample size. We also consider more granular levels of analysis, using fixed effects for county (columns 3–4) and enumeration-district (columns 5–6). Despite these changes, estimates are largely unaffected. Together, the results in panel (a) suggest that first-generation migrants not only transplant Confederate affinity for the KKK but also transmit that affinity to their sons in the next generation. Second, in panel (b), we find evidence consistent with horizontal transmission of KKK affinity from the Confederate diaspora to their neighbors without Southern heritage. In this case, we restrict the analysis to white men born outside the South and whose parents were also born outside the South. We then regress these men’s KKK membership indicator on measures of physical proximity to the Confederate diaspora. We find highly localized associations suggestive of neighborhood exposure effects. Non-Southern whites with next-door neighbors from the first- or second-generation Confederate diaspora are nearly 2 p.p., or 8%, more likely to be KKK members than those whose next-door neighbors have no Southern heritage (columns 1 and 3).32 We find similar estimates for Denver county and the broader metro area (columns 1–2 and 3–4, respectively), and for neighbors from the first- and secondgeneration (columns 2 and 4). Finally, we see, in columns 5–6, that the transmission of KKK activity extends beyond the family and next-door neighbors to others within the broader neighborhood: a 1 p.p. increase in the diaspora size in the enumeration district is associated with a 0.5–0.8 p.p. increase in KKK membership among non-Southern-heritage whites. Together, the results in panels (a) and (b) of Table 11 are consistent with cultural transmission from the Confederate diaspora to their neighbors without Southern heritage. However, they are not dispositive of one-way transmission. It is possible that the two groups simultaneously joined the KKK for correlated reasons unrelated to cultural transmission. It is also possible that transmission flowed from non-Southern to Southern whites. The results in panel (a) perhaps go against the latter: diaspora whites are significantly more likely to be KKK members, making it plausible that they are the ones leading the early mobilization and diffusion efforts in their communities.33 Such an interpretation is also consistent 31 This number falls to 12.8% when accounting for individuals in the Census with non-unique names matched to the KKK records. Our regression estimates in Table 11 are robust to two alternative approaches to dealing with these non-unique links across the two datasets. Appendix Table D.3 shows robustness to reweighting each of the many-to-one matches by the random match probability, and Appendix Table D.4 shows robustness to flexibly controlling for the frequency of matches. 32 We identify next-door neighbors based on the zigzag enumeration procedure, following Logan and Parman (2017). 33 Under the assumption that the estimates in panel (b) are causal, our findings in Table 11 imply that the Confederate diaspora 25 with the causal SSIV estimates at the county level in Table 3. Moreover, in Appendix E, we present complementary evidence of one-way transmission: among non-Southern white migrants, those exposed to a larger Confederate diaspora in their new communities are more likely to name their children after Confederate leaders. By comparing name choices across children born before and after their household migrated, these mover-based strategies can effectively isolate causal spillover effects of exposure to a given place or population therein (see Bazzi et al., forthcoming, for a related application). Overall, these results suggest that migrants transmitted Confederate culture to non-Southern populations during a period of resurgent Confederate pride in the early 20th century. The patterns of KKK membership in Table 11 resonate with Gregory (2005, p. 294): “[t]he 1920s Klan had not been dominated by diaspora Southerners, but it had depended upon them for early expansion and some of its leadership.” Our analysis above, and in Appendix E, provides a granular perspective on the sort of localized, contact-based transmission of culture underlying our broader county-level findings throughout the paper. Such transmission was essential to the persistence of Confederate memory and norms supporting racial inequity over the long run. In the next section, we explore another institutional pathway for persistence that would have reinforced the legacy of the initial postbellum migrants out of the South. 5.4 Intergenerational Occupational Persistence and Institutional Capture This section presents an important additional insight into the mechanisms for persistence of the Confederate legacy outside the South. We saw in Section 3.2 that Southern white migrants were much more likely than other groups to occupy positions of authority in the late 1800s. And the foregoing results in Section 5.2 showed how those occupational choices may have helped to entrench Confederate norms in nascent institutions outside the South. In Table 12, we show that the second-generation born in the diaspora reinforced the institutional capture by their parents in the first generation, thus helping to sustain those norms over the long run, as we showed in Section 6. In particular, we revisit the occupational sorting analysis from Table 1 and add an additional term distinguishing the second generation migrants, i.e., those born outside the South to Southern-born parents. The patterns are consistent across all subsamples, comparing to other out-of-state migrants (panel b) and restricting to Western (panel c) and non-Western counties (panel d). Much like their parents who left the South, those born in the diaspora are significantly overrepresented in positions of public authority (column 1), and plausibly for reasons that went beyond the income opportunities in such positions (column 2). They entered the legal system and policing at rates on par with or greater than their parents (columns 3–4). And although they were relatively less likely to become religious leaders (column 6), these second-generation migrants realized even greater entry into positions in public administration (column 5) and education (column 6) that their parents were no more likely to occupy than other residents. These results complement our previous findings on the sorting by former slaveholders and their kin into authority positions (see Appendix Table C.2 and the discussion in Section 3.2). Through these positions of governance and persuasion, the Confederate diaspora was able to grow its influence over time. As the first generation of migrants aged out of the workforce, their children filled explains as much as 30% of KKK membership identified in the 1920 Census for the Denver metro area. This goes well beyond the 9.2% share of the white male population comprised of first- and second-generation Southern migrants as of 1920 and is consistent with their over-representation in the KKK as well as sizable spillover effects on non-Southerners. The 30% estimate is based on the coefficients in column 5, and it remains large at around 22% when using the coefficients in Appendix Table D.3 based on the alternative reweighting approach to dealing with multiple matches. 26 their shoes in important roles in public life. From these authority positions, they could shape subsequent generations and ultimately ensure that the roots of Confederate ideology continued to run deep in many communities across America over the long run. We turn now to evidence of this enduring legacy. 6 The Long Shadow of the Confederate Diaspora The results thus far have established how the Confederate diaspora shaped the cultural and institutional foundations of racial inequity in the early 20th century. In this brief and final section, we show that the early influence of these migrants propagated over the long run with adverse consequences for the socioeconomic well-being of Blacks outside the South. Despite the racially progressive reforms of the Civil Rights era in the 1960s, racial disparities persist across many domains of social and economic life. Our findings below suggest that the Confederate diaspora contributed to some of these lasting disparities in the 21st century. Long after early migrants—both slaveholders and non-slaveholders—had passed, their influence persisted as culture and institutions mutually reinforced racial animus across generations. 6.1 Employment, Segregation, and Policing We use our county-level SSIV framework to identify the long-run impact of the Confederate diaspora on three important dimensions of structural racial inequality. We measure racial wage gaps using the complete-count Census in 1940. We capture Black residential segregation in 1940 and 2000 using the indices from Logan and Parman (2017) and Escarce et al. (2011), respectively. We also collect data on jail occupants by race for 2010–18 from the Vera Institute of Justice and police killings by race for 2013–20 from the Mapping Police Violence Database. Table 13 reveals persistent patterns of racial inequity in large-diaspora counties: reduced relative earnings for Blacks as of 1940 (columns 1–2), increased Black residential segregation in 1940 and 2000 (columns 3–6), and increased probabilities of Black incarceration and police-induced mortality in the 2000s (columns 7–10). These effect sizes are large throughout, consistent with a roughly 4–5% decrease in relative Black earnings, 8–13% increase in Black segregation, a 5% increase in Black incarceration, and a 24–26% increase in the probability of police killing, all relative to sample means. Note that some of the segregation impacts, both in 1940 and in 2000, may be an outcome of sundown towns changing the geography of race across towns within counties. To what extent are these findings, like those in the early 20th century, driven by the outsized influence of the former slaveholding elites within the Confederate diaspora, as seen in Section 5.2? In Appendix Table C.8, we replicate, to striking effect, the double SSIV from Table 8, with the five outcomes in Table 13. Much like the short-run influence of the diaspora, its enduring legacy is largely explained by former slaveholder migrants. Overall, these findings are consistent with recent work connecting Confederate memorialization to adverse racial outcomes (O’Connell et al., 2020; Rahnama, 2022; Williams, 2021). As debates over the removal of Confederate monuments and battle flag have raged in recent years, some have defended such symbols and memorials as celebrations of Southern pride and national heritage. Our findings are consistent with the view that this memorabilia may have acted as harbingers of more overt and harmful racial animus, insofar as their dissemination was often part of a sustained effort to preserve white supremacy and promote racial exclusion. 27 6.2 Racial Animus Black disadvantage in counties with a large Confederate diaspora may be reinforced, over the long run, by racial animosity among white residents. Using the General Social Survey (GSS) from 1993–2014 and the Cooperative Congressional Election Survey (CCES) from 2010–16, we track whites’ views on prominent racial issues. These surveys are rich in detail but limited in geographic representativeness, which hinders use of the SSIV strategy. Nevertheless, we find suggestive evidence that a larger Confederate diaspora of postbellum is associated, at the turn of the 21st century, with stronger preferences for racial separatism, white supremacist beliefs, and racially conservative political ideology. Appendix Table D.2 reports these estimates. A 1 p.p. increase in the migrant share in 1900 is associated with a 5.5% increase in mean preferences for white residential segregation, a 7.7% increase in pro-segregation beliefs, and an 11% increase in opposition to interracial marriage. Respondents are 0.5% more likely to believe Blacks must work their way up without government help, 0.6% more likely to state that systemic racism does not exist, and 1.4% more likely to express fear of other races.34 These patterns are consistent with a feedback loop between cultural, organizational, and institutional mechanisms identified earlier. As Confederate norms deepened and former slaveholders captured important institutions, Black populations fled from and later avoided many towns and communities across America. KKK mobilization, differential incarceration, and forced exclusion through sundown towns changed the geography of race. Together, these limited the scope for local interracial contact that might have otherwise ameliorated biases and animus among white Americans over the long run. 7 Discussion A debate has raged in recent years among the general American public as to the significance of Confederate memorials and iconography. Some claim that these serve merely as innocuous symbols of Southern pride and are part of America’s past and heritage. Others argue that they are harbingers of overt racial hate. These debates have generated intense public emotion, at times spilling over into violence. History and memory are at the heart of these debates, and as such require continued research to elucidate the origins and consequences of the Confederate legacy. This paper offers a new perspective. We showed how Confederate memorialization, together with more explicit expressions of white supremacy associated with the postbellum South, spread across the country in the early 20th century. Nearly one million Southern whites left the South in the few decades after the war and transmitted Confederate culture within non-Southern communities and through local institutions, over which they had an outsized influence. Such influence arose in part through sorting decisions among migrants—and especially former slaveholders—which favored destinations in the country’s developing West, as well as occupations with outward-facing public authority. Our empirical approach reinforces the work of historians on the Confederate legacy and provides novel insights on how the postbellum diaspora helped ensure, by the early 20th century, that “national reconciliation had been achieved on the South’s terms” (Cox, 2003). The diaspora’s early influence remained salient over the long run. We linked the historical presence of postbellum Southern white migrants to persistent racial inequities in labor and housing markets and in 34 See Chinoy et al. (2023) for more on the intergenerational implications of Confederate culture, based on our CCI measure. 28 policing practice. The visible legacy of this “Confederate diaspora” offers an important lesson for policymakers weighing the costs and benefits of Confederate cultural preservation. Ultimately, to advance racial progress in the U.S. means first understanding racism, something to which many Confederate cultural artifacts are incontrovertibly tied. We hope this paper stimulates future work on the deep roots of racial animus and, especially, the role of migration in both exacerbating and dampening such animus. The Confederate diaspora studied in this paper set the stage for a larger exodus of Southern whites during the Great Migration. Dwarfing the million or so migrants that left the South in the three decades after the Civil War, upwards of 20 million whites migrated North and West in the 20th century. This mass migration followed pathways paved by the postbellum movers on which this present study focuses. Yet as we show separately in Bazzi et al. (forthcoming), their scale and geographic scope meant that the later migrants had a profound impact on the trajectory of mainstream conservative politics across America. Importantly, these impacts of the “other Great Migration” are distinct from the earlier impact of the Confederate diaspora and the comparatively extreme ideology of former enslavers. Ultimately, to understand how migration affected cultural and political change in the U.S., it is important to understand the effects—both distinct and combined—of the many, inevitably interconnected migratory episodes in the country’s history. These include not only the postbellum Southern white migration in this paper and the later diaspora in Bazzi et al. (forthcoming), but also the European migration that shaped the South well before both emigration episodes (Grosjean, 2014), the more general process of frontier settlement during the 19th century (Bazzi et al., 2020), and the Great Migration of Blacks during the 20th century (Calderon et al., 2019; Fouka et al., 2022). Together, the grand story of migration has colored and can be expected to continue to color the cultural and political geography of the U.S. 29