Further Exploring the Southern Slaveholder Diaspora

C Further Exploring the Southern Slaveholder Diaspora In this Appendix, we further explore the subset of the Confederate diaspora that was made up of former slaveholders within the antebellum South, as well as their direct kin. We begin by comparing their spatial and occupational sorting patterns to others in the Confederate diaspora without such ties to the institution of slavery. We then consider how their occupation choices influenced the diffusion of Confederate culture and other outcomes, using several additional robustness exercises. We conclude this Appendix by contextualizing these sorting patterns, and the implications thereof, through several back-of-theenvelope estimates of the former slaveholder diaspora’s size and geographic scope. C.1 Characterizing the Southern Slaveholder Diaspora Our analysis in Section 5 highlights the role of former slaveholders within the Confederate diaspora to driving its overall influence. Who were these slaveholder migrants? We begin by showing, in Appendix Figure C.1, that former slaveholders were more strongly attached to Confederate culture than were other Southern white migrants without slaveholding experience in the South as of 1860. The graph provides further validation of the hypothesis that former slaveholders maintained stronger affinity with the Confederacy even after they left its erstwhile border into the rest of the U.S. The remainder of this appendix then expands upon our sorting analyses in Section 3 and Appendix A in an effort to further characterize former slaveholder migrants within the diaspora. children with Confederate leader names (per 1,000) Figure C.1: Confederate Leader Names among Children of Southern Former Slaveholders whites born to slaveholders whites born to non-slaveholders 5 4 3 2 1 0 1851-55 1856-60 1861-65 1866-70 1871-75 1876-80 Notes: Confederate leader name frequencies across five-year birth year intervals from 1801 to 1880 for children (ages 0–9) in the U.S. complete-count 1860, 1870, and 1880 Censuses among different subsets of Southern white fathers in the U.S. Census: those born to Southern white fathers who were matched to the 1860 Slave Schedule (i.e., former slaveholders) and those born to Southern white fathers who were not matched to the 1860 Slave Schedule and who were not matched to the same household as someone matched to the 1860 Slave Schedule (i.e., were not slaveholders or kin). The latter group of fathers (non-slaveholders) thus also includes Southern whites who were not successfully linked to the 1860 Census. An individual’s Confederate name match equals one if their given, first name is highly likely to have been given in reference to a Confederate leader. This includes individuals whose first name includes a leader’s full name (e.g., “Robert Lee”), as well as distinctive nicknames like Stonewall and last names like Beauregard. The list of Confederate leader names includes those with multiple public symbols in the Confederacy in the Southern Poverty Law Center’s (SPLC) “Whose Heritage?” Project. C-1 Destination Characteristics of Former Slaveholder Migrants. We examine former slaveholders’ typical destination choices, relative to the average postbellum Southern white migrant. To do this, we use linked Census records from the Census Linking Project (Abramitzky et al., 2020), which allow us to study sorting behavior among a subsample of Southern white migrants that can be linked back to the 1860 Census—and therefore potentially to the 1860 Slave Schedule. These linked Census records provide information on migrants’ county of origin and decade of migration. With this tracking ability, we can estimate the following equation, using a stacked sample of Southern white migrants for the Census periods through τ ∈ {1880, 1900}: yidoτ = α slaveholderidoτ + x0idoτ β + ηoτ + εidoτ , where yidoτ denotes some characteristic of destination county d of Southern white migrant i from origin county o in the Census period through τ . Using our baseline definition of former slaveholders based on listed slaveholders (panel a) and another that also includes their direct kin (panel b), Appendix Table C.1 shows that Southern slaveholder migrants, relative to non-slaveholders, gravitated towards more remote Western counties that looked more economically, agroclimatically, and politically similar to the South. These patterns speak to the possibility, raised in historical accounts, that former slaveholder migrants may have sought out areas where the socioeconomic hierarchies of the South might be replicated. We discuss these estimates in Section 3. Table C.1: Destination Characteristics of Former Slaveholder Migrants Dependent Variable: West (1) Log Population Density (2) Cotton County (3) % Vote Share for Breckinridge, 1860 (4) % Union Army Enlistment (5) % Slaves, 1860 (6) (a) Based on Listed Slaveholders Only Former Slaveholder Observations Non-slaveholder mean Former Slaveholder Observations Non-slaveholder mean ∗∗∗ 0.022 (0.008) 27,878 0.11 0.031∗∗∗ (0.005) 27,878 0.11 ∗ -0.115 (0.059) 0.082∗∗∗ (0.012) 1.223∗∗ (0.504) -2.213∗∗ (0.877) 27,370 27,878 23,533 25,721 3.81 0.51 17.84 34.36 (b) Also Based on Slaveholder Household Members -0.117∗∗∗ (0.038) 27,370 3.83 0.080∗∗∗ (0.008) 27,878 0.50 1.673∗∗∗ (0.309) 23,533 17.41 -1.856∗∗∗ (0.525) 25,721 34.63 1.754∗∗∗ (0.294) 26,436 4.22 2.246∗∗∗ (0.177) 26,436 3.84 Notes: Regressions of observable destination county characteristics among white male migrants tracked from Southern to non-Southern counties between 1870–1880 or 1880–1900 on a dummy for a migrant’s former slaveholding status as of 1860 (equal to one for about 6% of sample individuals in row i). The set of destination counties excludes Southern counties belonging to states of the former Confederacy or Oklahoma. The sample of former slaveholders is constructed from the 1860 U.S. Census Slave Schedule, which we match to the 1860 U.S. Census of Population. We then use linked Census records to track white male migrants from Southern to non-Southern counties across Census periods for decades following the Civil War from 1870–1900. For comparability, all individuals in the sample must be able to be matched to the 1860 Census. All regressions include origin county×year fixed effects. Individual controls include a cubic in age, marital status, and number of children. Robust standard errors in parentheses. Significance levels are denoted by * p < 0.10, ** p < 0.05, *** p < 0.01. Former Slaveholder Sorting into Positions of Authority. Occupational sorting patterns at destination likewise speak to the incentives and preferences of these former slaveholder migrants. For context, we begin by briefly reviewing the occupational sorting patterns of postbellum Southern white migrants more generally, before highlighting the distinct choices of former slaveholder migrants therein. Recall that Table 1 uses the 1900 complete-count U.S. Census to characterize occupational sorting patterns C-2 of white working-age men in the postbellum Southern white diaspora, relative to non-Southern whites. Panel (b) of Table 1 further limits the sample to out-of-state migrants, estimating the following equation: P(occupationid = occ) = α southerni + x0i β + θd + εid , where “occ” denotes various occupational choices. These include the following positions of authority: • Law enforcement: includes policemen and detectives (occ1950=773, occscore=32), marshals and constables (occ1950=771, occscore=24), and sheriffs and bailiffs (occ1950=782, occscore=30). In 1900, 37,811 men outside of the South were working as policemen or detectives, 6,570 as marshals or constables, and 5,827 as sheriffs or bailiffs, for a total of 50,208 men with an average occscore of 30.7. • Legal occupations: includes lawyers and judges (occ1950=55, occscore=62). In 1900, 83,301 men outside of the South were working as a lawyer or judge. • Public administration: includes officials and administrators (n.e.c.), public administration (occ1950 =250, occscore=36). In 1900, 20,530 men outside of the South were working as a public official or administrator. • Education: includes teachers (n.e.c.) (occ1950=93, occscore=27), college presidents and deans (occ1950=10, occscore=41), and professors and instructors (occ1950=12-29, occscore=41). In 1900, 106,208 men outside of the South were working as teachers, 197 as college presidents or deans, and 3,259 as college professors and instructors, for a total of 109,664 men with an average occscore of 27.4. • Religious occupations: includes clergyman (occ1950=9, occscore=24) and other religious workers (occ1950 =78, occscore=15). In 1900, 90,987 men outside of the South were working as a clergyman and another 2,108 as religious workers, for a total of 93,095 men with an average occscore of 23.8. These specifications (columns 3–7) control for relevant individual characteristics (xi : age, age2 , age3 , marital status, and number of children) and destination county fixed effects, θd . We document substantial occupational sorting toward positions of authority throughout Table 1 by postbellum Southern white migrants—for instance, into occupations of governance as well as religion—which we discuss at length in Section 3. We find similar patterns among migrants’ (non-Southern) children in Table 12. To understand the distinct role played by former slaveholder migrants (and kin) in driving these patterns, we must again turn to the subsample of migrants available through the linked Census records. We then estimate the following equation, using a stacked sample of Southern white migrants for the Census periods through τ ∈ {1880, 1900}: P(occupationidoτ = occ | southern white migrant) = α slaveholderidoτ + x0idoτ β + θdτ + ηoτ + εidoτ . Appendix Table C.2 shows that the former slaveholders were even more likely than non-slaveholders in the diaspora to sort into certain public-facing positions of authority, especially in law enforcement and public administration. C-3 Table C.2: Occupational Sorting by Former Slaveholders in the Confederate Diaspora Dependent Variable: Any Authority Position (1) (2) Lawyers or Judges (3) Law Enforcement (4) Public Administrators (5) Religious Workers (6) Educators (7) (a) Based on Listed Slaveholders Only Former Slaveholder ∗∗ 1.488 (0.663) ∗∗ 0.972 (0.457) 0.626∗ (0.361) 0.127 (0.153) 0.423∗ (0.254) 0.114 (0.324) 0.198 (0.252) Observations Non-slaveholder mean 23,362 2.50 23,362 2.50 23,362 0.76 23,362 0.26 23,362 0.15 23,362 0.91 23,362 0.43 (b) Also Based on Slaveholder Household Members Former Slaveholder Observations Non-slaveholder mean Occup. Income Category FE Origin County×Year FE Destination County×Year FE ∗∗∗ ∗ 1.084 (0.346) 0.441 (0.228) 0.774∗∗∗ (0.218) -0.016 (0.071) 0.177∗∗ (0.085) 0.052 (0.175) 0.097 (0.127) 23,362 2.33 23,362 2.33 23,362 0.62 23,362 0.27 23,362 0.15 23,362 0.87 23,362 0.42 No Yes Yes Yes Yes Yes – Yes Yes – Yes Yes – Yes Yes – Yes Yes – Yes Yes Notes: Regressions of various occupation indicators (×100) among white working-aged male migrants tracked from Southern to non-Southern counties between 1870–1880 or 1880–1900 on a dummy for a migrant’s former slaveholding status as of 1860 (equal to one for about 6% of sample individuals in row i). The “any authority” occupation indicator is based on any of the five positions of authority in columns 3–7. The set of destination counties excludes Southern counties belonging to states of the former Confederacy or Oklahoma. The sample of former slaveholders is constructed from the 1860 U.S. Census Slave Schedule, which we match to the 1860 U.S. Census of Population. We then use linked Census records to track white male migrants from Southern to non-Southern counties across Census periods for decades following the Civil War from 1870–1900. For comparability, all individuals in the sample must be able to be matched to the 1860 Census. All regressions include origin county×year and destination county×year fixed effects. Individual controls include a cubic in age, marital status, and number of children. Column 2 additionally controls for occupational income category fixed effects, based on bins of occscore width = 2. Occupational classifications follow those laid out by the Census Bureau’s 1950 (“occ1950”) definitions. Robust standard errors in parentheses. Significance levels are denoted by * p < 0.10, ** p < 0.05, *** p < 0.01. Beyond showing that Southern white migrants and especially former slaveholder migrants sorted strongly into positions of authority, we can also comment on whether this reflects a comparative advantage and/or taste for working in positions of authority, or is simply a byproduct of migrants’ elite socioeconomic status that traveled with them as they left the South (see Appendix Table C.3). Of course, many of these positions of authority happen to be characterized by high income. Given former slaveholders’ elite backgrounds in particular, it is possible that these sorting patterns simply reflect sorting into high-income occupations. C-4 Table C.3: Individual Characteristics of Migrants Versus Stayers Characteristic: All Linked Southern Whites Former Slaveholders Only Number Occ. Working in Occ. Working in Working in Working in Slaveholder of Slaves Income Position of Income Position of Agriculture Agriculture in 1860 in 1860 Authority Score Authority Score (1) (2) (3) (4) (5) (6) (7) (8) Outmigrant 0.961∗∗∗ (0.146) 0.363∗∗∗ (0.109) -8.410∗∗∗ (0.477) -0.045∗∗∗ (0.003) 1.199∗∗∗ (0.390) 0.969∗ (0.496) -6.257∗∗∗ (1.203) 0.080 (0.455) Observations 285,138 Outcome mean (non-migrant) 16.32 344,988 1.68 344,988 64.68 344,988 0.11 31,917 18.08 36,240 2.83 36,240 12.04 36,240 6.27 Notes: Regressions of various individual characteristics among white men living in the South in 1860 (columns 1–4) and among slaveholders living in the South in 1860 (columns 5—8) on whether they subsequently migrated to the non-South between 1870–1880 or 1880–1900. The sample of former slaveholders is constructed from the 1860 U.S. Census Slave Schedule, which we match to the 1860 U.S. Census of Population. We then use linked Census records to track which white male migrants moved from Southern to non-Southern counties across Census periods for decades following the Civil War from 1870–1900. For comparability, all individuals in the sample must be able to be matched to the 1860 Census. All regressions include origin county×year fixed effects. Individual controls include a cubic in age, marital status, and number of children. Robust standard errors in parentheses. Significance levels are denoted by * p < 0.10, ** p < 0.05, *** p < 0.01. We examine this possibility in two ways. First, columns 1 and 2 in Table 1 and Appendix Table C.2 group the five authority occupations into a single indicator. Column 2 additionally uses information on occupational income scores to compare occupational sorting patterns within occupational income categories. These estimates show sorting by Southern white migrants, and even more so by former slaveholder migrants, into positions of authority—even after controlling for income. Second, we run a large set of regressions using the specifications from panels (a) and (b) of each of these tables and then plot, in Appendix Figure C.2, the main regression coefficients against the occupational income score of the regressions’ given occupation. Across all of these, we fail to estimate a statistically significant correlation between differential sorting and occupational income (see the dashed regression line and corresponding estimates in bottom right of each graph panel). This further suggests that earnings potential alone is not driving sorting. For instance, sorting into the legal and religious sector is significantly stronger than sorting into a number of occupations of comparable earnings potential. Put differently, among occupations at similar levels of income, those with public-facing authority are more likely to attract Southern white migrants in the Confederate diaspora, and especially the former slaveholders therein. These migrants in the Confederate diaspora are differentially more likely to work in these authority positions than are non-Southerners, including migrants from elsewhere in the country. C-5 Figure C.2: Occupational Sorting: Authority Versus Income (a) Southern White Sorting (panel a of Table 1) (c) Slaveholder (b) Southern White Sorting (panel b of Table 1) Sorting (panel a of App. Table C.2) (d) Slaveholder Sorting (panel b of App. Table C.2) Notes: Triangles denote coefficients from regressions of occupations on a Southern-origin indicator among working-age white males outside the South (panels a and b) and a former slaveholder indicator among Southern white migrants outside the South (panels c and d). Details for all specifications correspond to those for columns 3—7 in panels (a and b) of Table 1 and panels (a and b) of Appendix Table C.2, respectively. See the notes to those tables for all details. Triangle darkness denotes significance level (i.e., light grey = p < 0.10, grey = p < 0.05, black = p < 0.01). C.2 The Distinct Influence of Former Slaveholder Migrants This section further illustrates the distinct influence of former slaveholders within the Confederate diaspora, through a series of robustness and placebo checks on Tables 9, 10, and 13. First, Appendix Tables C.4 and C.5 extend Table 9 to consider the relative importance of sorting into occupations of authority by postbellum Southern white migrants more generally. Appendix Table C.5 extends Appendix Table C.4 by controlling for (instrumented) slaveholder migrant shares. These results together further suggest the central role of former slaveholders in driving our county-level effects. Second, Appendix Table C.6 extends Table 9 through a set of placebo regressions, which consider the relative importance of sorting by these migrants into occupations of similar income and sector size to our positions of authority, but with no de jure component of power or influence. Of course, some of these more private-facing positions, like medical doctors (the placebo for lawyer/judge), may still entail considerable scope for public influence. As such, we interpret these placebos with a measure of caution. C-6 In what follows, we describe placebo occupation(s) for each authority occupation: • For law enforcement occupations: mechanics and repairmen (n.e.c.) (occ1950=554, occscore=30), funeral directors and embalmers (occ1950=54, occscore=31), and sports instructors and officials (occ1950=91, occscore=32). In 1900, 31,119 men outside of the South were working as mechanics or repairmen, 13,528 as funeral directors or embalmers, and 6,185 as sports instructors or sports officials, for a total of 50,832 men with an average occscore of 30.5. • For legal occupations: physicians and surgeons (occ1950=75, occscore=80). In 1900, 113,070 men outside of the South were working as physicians and surgeons. • For public administrative occupations: railroad switchmen (occ1950=681, occscore=36) and advertising agents and salesmen (occ1950=400, occscore=35). In 1900, 17,644 men outside of the South were working as switchmen and 3,021 working as advertising agents or salesman, for a total of 20,665 men with an average occscore of 35.9. • For education occupations: metal molders (occ1950=561, occscore=28), photographers (occ1950 =74, occscore=27); Glaziers (occ1950=530, occscore=28), and surveyors (occ1950=92, occscore =28). In 1900, 82,468 men outside of the South were working as metal molders, 21,587 as photgraphers, 2,407 as glaziers, and 3,745 as surveyors, for a total of 110,207 men with an average occscore of 27.8. • For religious occupations: textile weavers (occ1950=684, occscore=23), apprentice machinists and toolmakers (occ1950=604, occscore=24), and metalworking trades apprentices (n.e.c.) (occ1950 =612, occscore=23). In 1900, 77,079 men outside of the South were working as textile weavers, 9,579 as apprentice machinists or toolmakers, and 5,037 as metalworking trades apprentices, for a total of 92,325 men with an average occscore of 23.1. The first set of estimates in Appendix Tables C.6 reiterate those in Table 9 above, wherein local public officials and religious leaders with ties to slavery played a uniquely important role in diffusing Confederate culture and white supremacy, even beyond the baseline effect of former slaveholder migrants. In contrast, placebo occupations of comparable income but less public authority exhibit little such importance, with estimates that are smaller and statistically insignificant across specifications. This again suggests that occupational pathways to public influence were key to former slaveholder migrants’ outsized influence. Finally, Appendix Table C.8 augments the specifications in Table 13 to explore the long shadow of former slaveholders in shaping Black opportunities outside the South. SSIV estimates show that, as with the short-run influence of the Confederate diaspora, its enduring legacy is largely explained by former slaveholder migrants. C-7 Table C.4: Using All Linked Southern White Migrants for Table 9 Dependent Variable: (1) % Southern Whites, 1900 (6) Any Confederate Memorials (7) Any UDC Chapters (8) Any KKK Chapters (9) Any Lynchings of Blacks (10) 0.124 (0.038) 0.116∗∗∗ (0.040) 0.032∗∗∗ (0.010) 0.037∗∗∗ (0.013) 0.031∗ (0.016) 0.017∗ (0.010) 0.289∗∗∗ (0.103) 0.157 (0.172) 0.099 (0.186) 0.334∗∗ (0.149) 0.336∗∗∗ (0.091) 0.175∗ (0.100) -0.019 (0.061) 0.186 (0.134) 0.098 (0.068) 0.076∗ (0.040) 0.031 (0.051) 0.046 (0.062) -0.014 (0.158) 0.140∗∗ (0.063) 0.080∗∗ (0.031) 0.112∗∗∗ (0.040) 0.122∗ (0.072) 0.056 (0.234) 0.067 (0.063) 0.104∗∗ (0.043) 0.051 (0.047) 0.008 (0.050) -0.130∗∗ (0.053) 0.030 (0.047) 0.076∗∗ (0.031) -0.019 (0.027) IV Yes Yes Yes 1,701 0.76 14.7 0.00 0.00 IV Yes Yes Yes 1,701 0.76 15.0 0.00 0.00 IV Yes Yes Yes 1,701 0.25 15.0 0.00 0.00 IV Yes Yes Yes 1,701 0.09 15.0 0.00 0.00 IV Yes Yes Yes 1,701 0.37 15.0 0.01 0.00 IV Yes Yes Yes 1,701 0.05 15.0 0.03 0.00 Confederate Cultural Activity (CCI Score, from 0–4) ∗∗∗ 0.129 (0.039) (2) ∗∗∗ 0.130 (0.039) (3) ∗∗∗ 0.129 (0.040) (4) ∗∗∗ 0.120 (0.040) (5) ∗∗∗ Any Linked White . . . from South? Public Administrators 0.192 (0.179) Religious Workers 0.361 (0.260) 0.408∗∗ (0.159) Law Enforcement C-8 0.393∗∗∗ (0.090) Lawyers or Judges Educators Estimator State FE Baseline controls 1870 share control Observations Outcome mean F-statistic Anderson-Rubin, p-val KP Underident., p-val IV Yes Yes Yes 1,701 0.76 15.0 0.00 0.00 IV Yes Yes Yes 1,701 0.76 14.9 0.00 0.00 IV Yes Yes Yes 1,701 0.76 15.4 0.00 0.00 IV Yes Yes Yes 1,701 0.76 14.7 0.00 0.00 Notes: This table re-estimates Tables 9 excluding former slaveholder migrant shares and including county-level indicators for whether any Southern white male migrants (linked to a Southern household in 1860) is working in a given occupation in that non-Southern county in 1900. See the notes of Tables 8 for all other details on the specification. Excluded Southern counties are those belonging to states of the former Confederacy and Oklahoma. Standard errors are clustered at the 60×60 square-mile grid cell level, following the approach of Bester et al. (2011). Significance levels are denoted by * p < 0.10, ** p < 0.05, *** p < 0.01. Table C.5: Controlling for Former Slaveholder Migrants in Table C.4 (1) (2) (3) (4) (5) (6) Any Confederate Memorials (7) -0.043 (0.063) 2.703∗∗ (1.095) -0.042 (0.063) 2.690∗∗ (1.101) -0.043 (0.063) 2.704∗∗ (1.095) -0.048 (0.062) 2.708∗∗ (1.100) -0.042 (0.063) 2.691∗∗ (1.127) -0.049 (0.063) 2.742∗∗ (1.129) -0.016 (0.020) 0.777∗∗ (0.351) -0.015 (0.020) 0.871∗∗ (0.353) -0.001 (0.023) 0.570 (0.402) -0.016 (0.015) 0.524∗∗ (0.248) 0.014 (0.202) 0.120 (0.259) -0.036 (0.285) 0.352 (0.220) 0.209 (0.141) -0.073 (0.203) -0.029 (0.082) 0.149 (0.140) 0.103 (0.074) 0.040 (0.050) -0.040 (0.070) 0.034 (0.082) -0.056 (0.165) 0.146∗ (0.086) 0.039 (0.045) 0.033 (0.072) 0.113 (0.089) 0.025 (0.243) 0.067 (0.075) 0.079 (0.054) -0.001 (0.069) 0.002 (0.060) -0.155∗∗ (0.073) 0.036 (0.057) 0.051 (0.037) -0.066 (0.042) IV Yes Yes Yes 1,679 0.77 2.4 0.02 IV Yes Yes Yes 1,679 0.77 2.4 0.02 IV Yes Yes Yes 1,679 0.25 2.4 0.02 IV Yes Yes Yes 1,679 0.09 2.4 0.02 IV Yes Yes Yes 1,679 0.38 2.4 0.02 IV Yes Yes Yes 1,679 0.05 2.4 0.02 Dependent Variable: % Southern Whites, 1900 % Southern Former Slaveholders, 1900 Confederate Cultural Activity (CCI Score, from 0–4) Any UDC Chapters (8) Any KKK Chapters (9) Any Lynchings of Blacks (10) Any Linked White . . . from South? Public Administrators 0.145 (0.261) Religious Workers 0.148 (0.309) 0.375∗ (0.223) Law Enforcement Lawyers or Judges 0.227 (0.138) C-9 Educators Estimator State FE Baseline controls 1870 share controls Observations Outcome mean KP Joint F-statistic KP Underident., p-val IV Yes Yes Yes 1,679 0.77 2.6 0.01 IV Yes Yes Yes 1,679 0.77 2.5 0.01 IV Yes Yes Yes 1,679 0.77 2.5 0.01 IV Yes Yes Yes 1,679 0.77 2.5 0.01 Notes: This table re-estimates Tables 9 alongside county-level indicators, based on any Southern white (male) migrants linked to any Southern household (as of 1860) working in a given occupation in that non-Southern county (as of 1900). See the notes of Tables 8 for all other details on instruments and controls. Excluded Southern counties are those belonging to states of the former Confederacy and Oklahoma. Standard errors are clustered at the 60×60 square-mile grid cell level, following the approach of Bester et al. (2011). Significance levels are denoted by * p < 0.10, ** p < 0.05, *** p < 0.01. Table C.6: Placebo Exercise: Using Non-Authority Positions of Similar Income in Table 9 Dependent Variable: % Southern Whites, 1900 % Southern Former Slaveholders, 1900 (1) (2) (3) (4) -0.048 (0.064) 2.731∗∗ (1.105) -0.040 (0.062) 2.670∗∗ (1.087) -0.042 (0.063) 2.705∗∗ (1.102) -0.042 (0.063) 2.691∗∗ (1.097) Confederate Cultural Activity (CCI Score, from 0–4) (5) (6) (7) (8) (9) -0.041 (0.062) 2.675∗∗ (1.092) -0.042 (0.063) 2.690∗∗ (1.108) -0.042 (0.062) 2.648∗∗ (1.116) -0.044 (0.062) 2.719∗∗ (1.089) -0.041 (0.063) 2.729∗∗ (1.160) (10) (11) (12) (13) -0.041 (0.063) 2.674∗∗ (1.097) -0.047 (0.062) 2.724∗∗ (1.176) -0.042 (0.060) 2.676∗∗ (1.080) -0.046 (0.060) 2.720∗∗ (1.146) Any Linked . . . from Southern Slaveholder Households? Public Administrators 0.735∗∗∗ (0.227) Public Administrators (Placebo) 0.727∗∗∗ (0.237) 0.598 (0.505) 0.537 (0.544) 1.140∗∗∗ (0.182) Religious Workers Religious Workers (Placebo) 1.141∗∗∗ (0.182) 0.130 (0.364) Law Enforcement 0.110 (0.373) 0.192 (0.329) C-10 Law Enforcement (Placebo) 0.270 (0.354) 0.086 (0.448) Lawyers or Judges 0.018 (0.436) 0.222 (0.212) Lawyers or Judges (Placebo) 0.227 (0.214) 0.166 (0.237) Educators 0.144 (0.243) -0.160 (0.341) Educators (Placebo) Estimator State FE Baseline controls 1870 share controls Observations Outcome mean KP Joint F-statistic KP Underident., p-val IV Yes Yes Yes 1,679 0.77 2.5 0.01 IV Yes Yes Yes 1,679 0.77 2.6 0.01 IV Yes Yes Yes 1,679 0.77 2.5 0.01 IV Yes Yes Yes 1,679 0.77 2.5 0.01 IV Yes Yes Yes 1,679 0.77 2.5 0.01 IV Yes Yes Yes 1,679 0.77 2.5 0.01 IV Yes Yes Yes 1,679 0.77 2.5 0.02 IV Yes Yes Yes 1,679 0.77 2.6 0.01 IV Yes Yes Yes 1,679 0.77 2.4 0.02 -0.258 (0.345) 0.227 (0.250) IV Yes Yes Yes 1,679 0.77 2.5 0.01 IV Yes Yes Yes 1,679 0.77 2.3 0.02 0.168 (0.246) IV Yes Yes Yes 1,679 0.77 2.6 0.01 0.705∗∗∗ (0.249) 0.531 (0.571) 1.128∗∗∗ (0.181) 0.096 (0.392) 0.189 (0.374) 0.034 (0.424) 0.184 (0.206) 0.095 (0.247) -0.265 (0.333) 0.121 (0.245) IV Yes Yes Yes 1,679 0.77 2.4 0.02 Notes: This table re-estimates Tables 9 alongside county-level indicators, based on any Southern white (male) migrants linked to former slaveholder households (as of 1860) working in a given occupation in that nonSouthern county (as of 1900). Placebo categories are constructed to be of similar size and occupational income score (occscore) to their comparison category, while lacking the authority component. See the notes of Tables 8 for all other details on instruments and controls. Excluded Southern counties are those belonging to states of the former Confederacy and Oklahoma. Standard errors are clustered at the 60×60 square-mile grid cell level, following the approach of Bester et al. (2011). Significance levels are denoted by * p < 0.10, ** p < 0.05, *** p < 0.01. Table C.7: Adding Respondents’ Contemporaneous 1920 County FE to Table 10 (1) Black ∗∗∗ 0.254 (0.031) Dependent Variable: Person is Incarcerated in 1920 (2) (3) (4) (5) ∗∗∗ 0.287 (0.035) ∗∗∗ 0.244 (0.031) ∗∗∗ 0.290 (0.037) ∗∗∗ (6) 0.297 (0.050) 0.327∗∗∗ (0.052) -0.191∗ (0.103) 0.319∗∗ (0.143) 0.404∗∗∗ (0.084) Any . . . from Southern Slaveholder Households in 1900 County of Residence? Law Enforcement Law Enforcement × Black 0.012 (0.025) 0.337∗∗ (0.149) 0.330∗∗ (0.159) Lawyers or Judges 0.016 (0.012) 0.213∗ (0.113) Lawyers or Judges × Black 0.175 (0.123) Public Administrators Public Administrators × Black Estimator Initial state FE Initial county FE Destination county FE Observations Outcome mean (whites) OLS Yes Yes 4,227,990 0.06 OLS Yes Yes 4,227,990 0.06 OLS Yes Yes 4,227,990 0.06 OLS Yes Yes 4,227,990 0.06 OLS Yes Yes 4,227,990 0.06 OLS Yes Yes 4,227,990 0.06 Notes: This table re-estimates Table 10 including an additional set of fixed effects for the respondents’ contemporaneous county in 1920. The dependent variable is a binary indicator (×100). Standard errors are clustered by initial county. Significance levels are denoted by * p < 0.10, ** p < 0.05, *** p < 0.01. C-11 Table C.8: Using Southern Former Slaveholders for Table 13 (1) (2) (3) (4) (5) (6) (7) (a) Dep. Var.: Black to Non-Black Earnings Ratio, 1940 % Southern Whites, 1900 ∗∗∗ -0.039 (0.014) -0.393∗∗ (0.189) % Southern Former Slaveholders, 1900 -0.423∗∗ (0.191) 0.028 (0.042) -0.978 (0.663) 0.024 (0.042) -0.932 (0.673) 0.020 (0.031) -0.834∗ (0.461) 0.038 (0.059) -1.040 (0.954) (b) Dep. Var.: Black Residential Segregation, 1940 % Southern Whites, 1900 ∗∗ 0.007 (0.003) 0.079∗∗ (0.037) % Southern Former Slaveholders, 1900 0.081∗∗ (0.036) -0.008 (0.006) 0.234∗∗ (0.118) -0.009 (0.007) 0.241∗ (0.126) -0.007 (0.005) 0.214∗∗ (0.097) -0.007 (0.007) 0.204∗ (0.118) (c) Dep. Var.: Black Residential Segregation, 2000 % Southern Whites, 1900 ∗∗∗ 0.009 (0.003) 0.105∗∗∗ (0.030) % Southern Former Slaveholders, 1900 0.110∗∗∗ (0.031) -0.005 (0.005) 0.223∗∗ (0.091) -0.005 (0.006) 0.219∗∗ (0.095) -0.004 (0.004) 0.204∗∗∗ (0.076) -0.001 (0.005) 0.154∗ (0.080) (d) Dep. Var.: Black County Jail Shares, 2010–18 % Southern Whites, 1900 ∗∗ 0.757 (0.313) 6.643∗ (3.763) % Southern Former Slaveholders, 1900 8.054∗∗ (3.809) -1.404 (0.997) 34.501∗∗ (16.806) -1.316 (1.020) 33.249∗ (17.183) -1.054 (0.699) 29.190∗∗ (12.473) -1.415 (1.131) 33.843∗ (18.152) (e) Dep. Var.: Any Police Killings of Blacks, 2013–20 % Southern Whites, 1900 0.034∗∗ (0.013) 0.010 (0.014) 0.424 (0.274) 0.008 (0.012) 0.454∗ (0.268) 0.023∗ (0.013) 0.255 (0.269) IV Yes Yes Yes Yes 0.299∗∗ (0.127) 0.298∗∗ (0.125) IV Yes Yes Yes IV Yes Yes Yes IV Yes Yes Yes Yes IV Yes Yes Yes IV Yes Yes Yes Yes IV Yes Yes Yes Yes Yes 1,701 15.0 0.00 1,679 19.8 0.00 1,679 20.2 0.00 1,679 2.5 0.01 7.1 0.01 5.1 0.02 1,679 2.4 0.02 6.5 0.01 4.8 0.03 1,679 3.9 0.00 6.5 0.01 4.8 0.03 % Southern Former Slaveholders, 1900 Estimator State FE Baseline controls 1870 share control(s) % Non-Southern slaveholders control Linked Census measures only? Sorting controls Observations KP Joint F-statistic KP Joint Underident., p-val % S. Whites, SW F-statistic % S. Whites, Underident., p-val % Slaveholders, SW F-statistic % Slaveholders, Underident., p-val 0.010 (0.013) 0.413 (0.257) Yes 1,677 2.6 0.01 6.8 0.01 5.2 0.02 Notes: This table re-estimates Table 13 using the same set of specifications, across columns, as Table 8. See the notes therein for details. Standard errors are clustered at the 60×60 square-mile grid cell level, following the approach of Bester et al. (2011). Significance levels are denoted by * p < 0.10, ** p < 0.05, *** p < 0.01. C-12 C.3 The Size and Scope of the Southern Slaveholder Diaspora In this section, we use the linked Census data—in combination with complete-count Census data and estimates from this Appendix—to further characterize the size and geographic spread of the former slaveholder diaspora. We develop several back-of-the-envelope estimates, both of the number of former slaveholder migrants from the South between 1870 and 1900 and of their entry into key positions of authority outside the South. These estimates help clarify the magnitude of sorting among former slaveholders. They also illustrate just how pervasive former slaveholders were among the Confederate diaspora, and how overrepresented they were in positions of influence within their new communities. Together, these results provide useful context for the effects observed, for instance, in Tables 8 and 9. How Many Southern (Slaveholder) Migrants Were There? Although it is difficult to quantify precisely how many Southern whites, and former slaveholders in particular, left the former Confederacy for the rest of the country in the three decades after the Civil War, we aim to impute these flows using data from the complete-count and linked Census records from the Census Linking Project (Abramitzky et al., 2020). Based on these, we estimate at least 921,119 Southern white migrants from 1870–1900, including 61,430 former slaveholders and an additional 127,696 of their household kin. These counts are constructed as follows: for each of these three groups, we calculate an outmigration rate between 0 and 1 for each Census period through τ ∈ {1880, 1900} from each Southern county. We then multiply these by a measure of the relevant origin county population from the previous Census period τ − 1. Concretely, to estimate total Southern white migrant flows, we first use the linked Census records together with the complete-count Census to approximate the number of Southern white migrants from each origin county to all non-Southern counties for each Census period, based on equation (B.1). Aggregating across all Southern origin county-years, this gives us an estimate of the total Southern white migrant population from all Southern counties to all non-Southern counties, of 921,119 individuals. We repeat the same process for former slaveholder migrants, using the share of linked former slaveholders in a given Southern county in a given Census year to impute the initial former slaveholder population from which migrants were sourced. Concretely, we use the linked Census records together with the complete-count Census to approximate, for each Census period, total Southern former slaveholder outmigration flows from o to all non-Southern counties, based on equation (B.2). Aggregating across all Southern origin county-years, this gives us a count estimate of the total former slaveholder migrant population from all Southern counties to all non-Southern counties, of 61,430 individuals. Note that while the first two terms in equation (B.2) are constructed using only the available data on linked men, our final approximation is gender inclusive. This is based on the fact that (i) some slaveholders in the 1860 Slave Schedule were women, and (ii) slaves were implicitly the property of both a listed slaveholder and their spouse. This choice is supported by the fact that the share of all male Southern white migrants that were slaveholders in our linked samples, as used for Appendix Tables C.1 and C.2, is about 6%, which is similar to the share implied by the counts here. We also construct an even broader measure of former slaveholders that also includes direct household kin. In this measure, the first two terms in equation (B.2) further include male household members of a listed slaveholder, e.g., sons. This imputation produces a count estimate of the total migrant population from former slaveholder households from all Southern counties to all non-Southern counties, of 189,126 individuals. Note that because Census linking is imperfect and generally limited to men, actual counts are likely C-13 to differ from these. At the same time, use of linked Census records for the purposes of constructing aggregate data is shown to be accurate, for a slightly later period, in the parallel work of Bazzi et al. (forthcoming). Insofar as these estimates are accurate, it would imply that former slaveholders made up about 6.7% of the Confederate diaspora formed over 1870–1900, with their household kin making up an additional 13.9% of the diaspora. How Pervasive were Former Slaveholder Migrants in Positions of Authority? Although former slaveholders were somewhat rare within the Confederate diaspora—and even more so overall—our analyses throughout Section 3 and in the Appendix thus far suggest that former slaveholder migrants especially sorted into nascent communities in the West and, once there, sorted strongly into positions of authority and influence. As a result of their overrepresentation in positions of authority, especially where culture and institutions were incipient, former slaveholder migrants were able to play a distinct and outsized role in spreading Confederate culture and in entrenching it in local institutions. We now put these occupational sorting patterns into more concrete perspective, in order to further illustrate how the slaveholder diaspora could have had such large effects despite its small scale. We construct an additional set of back-of-the-envelope estimates for the purposes of examining the extent of local entry into key occupations, relative to local occupational sector size. We construct these estimates as follows. First, because use of the linked Census is limited when it comes to linking migrant individuals within narrow occupational categories across Censuses, we instead use our estimates from Appendix Table C.2 to back out shares of all former slaveholder migrants working in a given occupation. We do this for all sectors estimated as relevant in Tables 9 and/or 10, which includes our various governance (law enforcement, lawyers and judges, and public administration) and religious occupations—all of which demonstrated disproportionate sorting by Southern former slaveholders relative to non-Southern whites (see Table 1 and Appendix Table C.2 together). Second, we calculate for each non-Southern county d the total number of former slaveholder migrants in a given occupation. As occupations of authority were generally dominated by white men in 1900, we estimate former slaveholder migrants counts conservatively using total white male populations in 1900 from the complete-count Census, multiplied by former slaveholder migrant shares calculated from the linked Census records:   # slaveholders in d in 1900 linked to South in 1860 former slaveholderd,1900 = × white mend,1900 , # white men in d in 1900 linked to 1860 which we then multiply by the conditional means implied by the estimates in Appendix Table C.2 to generate counts for both governance occupations and religious occupations. Lastly, we divide these counts by the total number of workers in a given sector in each non-Southern county d, again from the 1900 complete-count U.S. Census. These shares, calculated for all counties in the conterminous non-Southern U.S., are plotted across the four panels of Appendix Figure C.3, which illustrates the relative prevalence and spatial reach of these former slaveholder migrants. This visual clarifies why we observe such large effects associated with former slaveholder migrants throughout Tables 8–10. Even in 1900, over three decades after the Civil War had ended, several nonSouthern counties are estimated as having had over half of their governance sector staffed by those with personal slaveholding backgrounds in the antebellum South. When we also include household kin of said slaveholders (e.g., sons), a nontrivial share of counties throughout the West and former border C-14 Figure C.3: The Spatial Distribution of Southern Slaveholders in Positions of Authority, 1900 (a) Slaveholders in Governance (c) Slaveholders (b) Slaveholders in Religion (+ Kin) in Governance (d) Slaveholders (+ Kin) in Religion Notes: Maps show estimated spatial diffusion across counties of former slaveholder migrants (excluding and including their direct kin) in occupations of governance (law enforcement, legal occupations, and public administration) and religion (religious clergy and other workers). states show a majority of their public employees as coming from slaveholding backgrounds. Although we do not use such imputations in our actual analysis, these back-of-the-envelope estimates nonetheless help to illustrate the likely reach of former slaveholders, given the sorting patterns we observe. Indeed, Southern white migrants are about twice as likely to have worked as a lawyer or judge as non-Southern white migrants in Table 1, and former slaveholder migrants are about twice as likely to have worked as a lawyer or judge as a non-slaveholding Southern white migrant in Table C.2. Such extreme sorting into such small sectors has the potential to generate significant overrepresentation, as seen here. C-15