Additional Tables and Figures

D D.1 Additional Tables and Figures Additional Tables Table D.1: Summary Statistics, Non-Southern County-Level Data Obs. Mean St. dev. Min. Max. Outcome variables Any Confederate memorials (e.g., monuments, place names)? Any United Daughters of the Confederacy chapters (1900–20)? Any 2nd Ku Klux Klan chapters (1915–40)? Any lynchings of Blacks occurred (post-1900)? Confederate cultural index (i.e., the sum of the above) Black to non-Black earnings ratio (1940) Black residential isolation index (1940) Black residential isolation index (2000) Black county jail shares (2010–18) Any police killings of Blacks (2013–20) 1,702 1,702 1,702 1,702 1,702 1,047 1,696 1,698 1,605 1,702 0.25 0.09 0.37 0.05 0.76 0.77 0.08 0.07 13.83 0.14 0.43 0.29 0.48 0.22 0.89 0.36 0.14 0.13 15.23 0.35 0.00 0.00 0.00 0.00 0.00 0.03 0.00 0.00 0.00 0.00 1.00 1.00 1.00 1.00 4.00 4.27 1.00 0.86 91.34 1.00 Explanatory variables % Southern-born whites (1900) Predicted % Southern-born whites (for 1870–1900) % Southern former slaveholders (1900) Predicted % Southern former slaveholders (for 1870–1900) 1,702 1,702 1,702 1,702 2.19 6.72 0.26 0.47 3.68 11.24 0.52 0.76 0.00 0.00 0.00 0.00 49.69 91.40 6.15 6.11 Controls Log population (1870) Log county area % Black population (1870) On frontier in 1860 Never on frontier (settled by 1790) Male Union Army enlistment rate during Civil War Male Union Army mortality rate during Civil War % Votes for Breckinridge (1860) % Slaves (1860) Cotton potential Tobacco potential Agricultural potential % Southern-born whites (1870) % Southern former slaveholders (1870) 1,702 1,702 1,702 1,702 1,702 1,702 1,702 1,702 1,702 1,700 1,700 1,700 1,702 1,680 8.39 6.59 2.13 0.40 0.10 25.64 3.21 8.53 2.13 0.20 0.58 0.45 5.70 0.28 2.18 0.86 5.93 0.49 0.30 27.88 4.91 15.59 6.75 0.27 0.35 0.21 11.18 0.58 0.00 3.14 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 13.74 9.91 68.75 1.00 1.00 100.00 87.50 79.81 58.44 0.74 0.97 0.68 91.40 10.00 Appendix Table D.1 shows summary statistics for outcome, explanatory, and control variables used in our county-level analysis. Boundaries for county-level variables are standardized to 2010 boundaries using ArcGIS, based on the procedure described in Ferrara et al. (2021). This lets us consistently match counties across time with historical Census data and prevents issues associated with the merging and splitting of counties across Census periods. This process involves creating unique units (henceforth county parts), based on where historical and 2010 counties intersect. We then calculate areas for each county part. We divide these by total (2010) county area to generate an area share-based weight. We then interpolate values of historical count variables for county parts based on these shares. Finally, these approximated counts are aggregated within each 2010 county. Counties get dropped from the analysis if any county part was not yet incorporated in the base year (e.g., as of 1870) and as such had no count data from which to derive harmonized values. D-1 Table D.2: The Legacy of the Confederate Diaspora for Racial Attitudes Dependent Variable: % Southern Whites, 1900 Estimator Survey State FE Survey wave FE Baseline controls Respondent controls Observations Outcome mean Adj. R2 Want Only Whites in Neighborhood (1) Whites Have Right to Segregate (2) Against Interracial Marriage (3) Fear Other Races (4) Blacks Must Work Their Way Up (5) No Systemic Racism Exists (6) 0.020∗∗∗ (0.007) 0.010∗ (0.005) 0.011∗∗ (0.004) 0.002∗ (0.001) 0.003∗∗∗ (0.001) 0.003∗∗ (0.001) OLS GSS Yes Yes Yes Yes 13,109 0.37 0.12 OLS GSS Yes Yes Yes Yes 1,803 0.13 0.05 OLS GSS Yes Yes Yes Yes 4,653 0.10 0.11 OLS CCES Yes – Yes Yes 27,882 0.14 0.01 OLS CCES Yes Yes Yes Yes 82,695 0.63 0.03 OLS CCES Yes Yes Yes Yes 82,676 0.50 0.03 Notes: Regressions of six survey indicators of racial sentiment from the GSS (column 1–3) and CCES (4–6) on the share of Southern whites in 1900 in non-Southern counties. Excluded Southern counties are those belonging to states of the former Confederacy and Oklahoma. Baseline controls include log county population in 1870 and log county area (in square miles), entered flexibly using quadratic terms. Respondent controls include sex and age, the latter entered flexibly using a quadratic term. All regressions include state fixed effects. 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. D-2 Table D.3: Reweighting the Estimates in Table 11 by Name Frequency Dependent Variable: Matched to KKK Member Records (2) (3) (4) (5) (1) (6) (a) All White Men Southern-Born Second-Generation Non-Southern-Born w/ Southern Parent ∗∗∗ 0.020 (0.004) 0.020 (0.004) 0.023∗∗∗ (0.005) 0.013∗∗∗ (0.003) Non-Southern-Born w/ Southern Mother – 128,721 0.10 0.013∗∗∗ (0.003) 0.015∗∗∗ (0.003) 0.014∗∗ (0.007) 0.020∗∗∗ (0.007) Non-Southern-Born w/ Southern Father Fixed Effects Observations Outcome mean (control) ∗∗∗ – 128,721 0.10 0.012∗∗∗ (0.003) 0.011∗∗∗ (0.003) 0.012∗∗∗ (0.004) 0.012∗∗∗ (0.004) County 240,628 0.13 0.012∗∗∗ (0.003) County 240,628 0.13 0.008∗∗ (0.004) 0.010∗∗∗ (0.004) District 240,627 0.13 District 240,627 0.13 (b) Only White Men with no Southern Heritage/Parentage Neighbors in Houses Next Door 1st or 2nd Gen. Southern White Neighbor 0.010∗∗∗ (0.003) 1st Gen. Southern White Neighbor 0.006∗∗∗ (0.002) 0.004∗ (0.002) 0.008∗∗∗ (0.002) 0.006 (0.004) 0.015∗∗∗ (0.005) 2nd Gen. Southern White Neighbor 0.002 (0.002) Neighbors in Enumeration District % 1st Gen. Southern Whites in District % 2nd Gen. Southern Whites in District Fixed Effects Observations Outcome mean (control) Estimator Sample Counties 0.001 (0.002) 0.004 (0.002) 0.001∗∗∗ (0.000) 0.004∗∗∗ (0.000) 0.001∗∗∗ (0.000) 0.004∗∗∗ (0.000) – 93,585 0.11 – 93,585 0.11 County 189,071 0.07 County 189,071 0.07 County 189,071 0.07 County 189,071 0.07 OLS Denver only OLS Denver only OLS All metro OLS All metro OLS All metro OLS All metro Notes: This table re-estimates Table 11 reweighting each observation by 1/N where N is the number of Census matches for the given individual’s first name–surname combination to the KKK ledger. The dependent variable is a binary indicator for whether a white male in the Denver, CO metropolitan area as of the 1920 U.S. Census can be found in Denver KKK membership records from the 1920s. Linking based on first and last names, using the ABE algorithm from Abramitzky et al. (2019) with NYSIIS standardized names. In panel (a), the sample includes all white men in the given area, and the regressors include indicators for whether they were born in the South and whether they were born outside the South but their parents were born in the South. In panel (b), the sample includes only those born outside the South to parents who were also born outside the South, and the regressors include an indicator for whether one’s next-door neighbors have first- and/or secondgeneration Southern white migrants and the share of first- and/or- second-generation Southern white migrants in one’s enumeration district. The latter is leave-out, excluding one’s next-door neighbors from the district total. There are 313 enumeration districts in Denver county and 527 in the greater metro area, which spans 14 counties. All regressions include controls for a cubic polynomial in age, marital status, and number of children. Standard errors are clustered by enumeration district. Significance levels are denoted by * p < 0.10, ** p < 0.05, *** p < 0.01. D-3 Table D.4: Adjusting the Estimates in Table 11 by Name Frequency Dependent Variable: Matched to KKK Member Records (2) (3) (4) (5) (1) (6) (a) All White Men Southern-Born Second-Generation Non-Southern-Born w/ Southern Parent ∗∗∗ 0.028 (0.005) 0.028 (0.005) 0.021∗∗∗ (0.005) 0.024∗∗∗ (0.003) Non-Southern-Born w/ Southern Mother – 129,248 0.24 0.024∗∗∗ (0.003) 0.019∗∗∗ (0.003) 0.011∗ (0.007) 0.018∗∗ (0.007) Non-Southern-Born w/ Southern Father Fixed Effects Observations Outcome mean (control) ∗∗∗ – 129,248 0.24 0.022∗∗∗ (0.003) 0.014∗∗∗ (0.003) 0.014∗∗∗ (0.004) 0.014∗∗∗ (0.004) County 241,298 0.21 0.022∗∗∗ (0.003) County 241,298 0.21 0.010∗∗ (0.004) 0.011∗∗∗ (0.004) District 241,297 0.21 District 241,297 0.21 (b) Only White Men with no Southern Heritage/Parentage Neighbors in Houses Next Door 1st or 2nd Gen. Southern White Neighbor 0.013∗∗∗ (0.004) 0.010∗∗∗ (0.003) 0.012∗∗ (0.005) 0.013∗∗ (0.006) 1st Gen. Southern White Neighbor 2nd Gen. Southern White Neighbor 0.004∗ (0.002) 0.007∗∗ (0.003) 0.011∗∗∗ (0.003) Neighbors in Enumeration District % 1st Gen. Southern Whites in District % 2nd Gen. Southern Whites in District Fixed Effects Observations Outcome mean (control) Estimator Sample Counties 0.003 (0.003) 0.005∗ (0.003) 0.002∗∗∗ (0.001) 0.005∗∗∗ (0.001) 0.002∗∗∗ (0.001) 0.005∗∗∗ (0.001) – 93,654 0.25 – 93,654 0.25 County 189,263 0.22 County 189,263 0.22 County 189,263 0.22 County 189,263 0.22 OLS Denver only OLS Denver only OLS All metro OLS All metro OLS All metro OLS All metro Notes: This table re-estimates Table 11 including indicators for deciles of the frequency N capturing the number of Census matches for the given individual’s first name–surname combination to the KKK ledger. The dependent variable is a binary indicator for whether a white male in the Denver, CO metropolitan area as of the 1920 U.S. Census can be found in Denver KKK membership records from the 1920s. Linking based on first and last names, using the ABE algorithm from Abramitzky et al. (2019) with NYSIIS standardized names. In panel (a), the sample includes all white men in the given area, and the regressors include indicators for whether they were born in the South and whether they were born outside the South but their parents were born in the South. In panel (b), the sample includes only those born outside the South to parents who were also born outside the South, and the regressors include an indicator for whether one’s next-door neighbors have first- and/or second-generation Southern white migrants and the share of first- and/or- second-generation Southern white migrants in one’s enumeration district. The latter is leave-out, excluding one’s next-door neighbors from the district total. There are 313 enumeration districts in Denver county and 527 in the greater metro area, which spans 14 counties. All regressions include controls for a cubic polynomial in age, marital status, and number of children. Standard errors are clustered by enumeration district. Significance levels are denoted by * p < 0.10, ** p < 0.05, *** p < 0.01. D-4 D.2 Additional Figures children with Confederate leader names (per 1,000) Figure D.1: Confederate Leader Names among 2nd Generation Southern and Non-Southern Whites .001 whites born in South to Southern white(s) whites born outside South to Southern white(s) whites born outside South to non-Southern whites .0008 .0006 .0004 .0002 0 1880 1890 1900 1910 birth year 1920 1930 1940 Notes: Three-year moving average in Confederate leader name frequencies across birth years among different subsets of second-generation Southern migrant (i.e., at least one parent born in the South) and non-Southern white populations (ages 0–9) in the complete-count U.S. Census by birth cohort year. See the notes to Figure 2 for details on the name measures. Figure D.2: Confederate Cultural Activity In and Outside the South (a) Any (c) Any (b) Any UDC Chapters Lynchings of Black People Confederate Memorials (d) Any KKK Chapters Notes: Maps show the diffusion of our four main county-level outcomes out of the South. The map for UDC chapters (panel a) is based on confirmed chapters during the 1900–20 period. The map for Confederate memorials (panel b) combines information on Confederate monuments as well as Confederate-inspired location names (i.e., places, streets, schools) matched from a restrictive set of name inputs, as previously used in panel b of Figure 2. The map for Black lynchings (panel c) is based on confirmed lynchings of Black people during the specified periods in the legend and features time variation. The map for KKK chapters (panel d) is based on the diffusion of chapters with time information. D-5 Figure D.3: Visualizing the “Great Retreat”, Black Depopulation in [. . . ] Outside the South Towns .75 .25 .7 .2 .65 .15 .6 .1 .55 .05 .5 % of towns with Black depopulation % All-white towns among all towns (a) All 0 1870 1880 1890 1900 1910 1920 1930 1940 % All-white towns (among all towns) % All-white towns (w/ >25 Black pop in 1870) % Towns w/ 1 Black family or less (w/ >25 Black pop in 1870) w/ ≥1,000 Pop. in 1870 .55 .25 .5 .2 .45 .15 .4 .1 .35 .05 .3 % of towns with Black depopulation % All-white towns among all towns (b) Towns 0 1870 1880 1890 1900 1910 1920 1930 1940 % All-white towns (among all towns) % All-white towns (w/ >25 Black pop in 1870) % Towns w/ 1 Black family or less (w/ >25 Black pop in 1870) Notes: This figure shows the share of non-Southern towns with all-white populations (left y-axis), and shares of non-Southern towns that had more than 25 Black residents as of 1870 with (i) no Blacks or (ii) with one Black family or less (right y-axis). The sample in (a) consists of all towns outside the former Confederacy and Oklahoma and in (b) is further limited to towns with populations of over 1,000 as of 1870. A version of this figure was previously featured in an earlier version of Bazzi et al. (forthcoming) and is based on a similar figure featured in Bazzi et al. (2022). D-6