Did Military Service in the Vietnam War Increase Racial Integration?
Does being drafted into an integrated military bring Americans of different races closer together years later? Bleemer (2026) studies men born in 1950–1952 who faced the Vietnam draft lottery, using 2022 voter records on about 3.6 million men. Among Black and Native American men, being drafted raised the chance of having a white spouse by 3.6 percentage points, from about 8%, and the chance of registering Republican by 2.9 points, from about 7%. They also likely lived in neighborhoods with about 2 points fewer minority residents. White veterans showed no such changes. Military service may explain about 20% of Black-white marriages in these birth cohorts.
Why This Article Was Selected for The Policy Scientist
Racial separation in marriage, neighborhoods, and political affiliation remains among the most durable features of American life, and it bears on who accumulates wealth, whose interests parties represent, and how citizens regard one another. Yet governments control few institutions that place young adults from different backgrounds in sustained contact: schools, public housing, and national service. That last option has regained salience as several European governments expand or reinstate military service, while diversity programs in American universities and the armed forces face legal and political challenge. Research since Allport (1954) has focused largely on short-run attitudes. This study observes behavior five decades later, using a strong draft-lottery design and a national voter file, though race is largely imputed and wartime conscription may not generalize. Bleemer has published influentially on race and college access. The study is a National Bureau of Economic Research working paper, a prominent series that lacks journal peer review.
Full Citation and Link to Article
Bleemer, Z. (2026). The Vietnam War and racial integration (NBER Working Paper No. 34900). National Bureau of Economic Research.
Central Research Question
Does extended service in a racially integrated national institution change how young adults live decades later? The study asks whether conscription into the U.S. military during the Vietnam War affected the later-life racial integration and political identity of Black, Native American, and white men. Integration is measured through marriage to a spouse of another race and the racial composition of the neighborhood where a man lives. The study also asks whether any effects were symmetric across minority and white veterans, and whether improved economic standing may help explain them.
Previous Literature
The study builds on intergroup contact theory, associated with Williams (1948) and Allport (1954). Paluck et al. (2019) conclude that transitory contact reduces prejudice in the short run among majority-group youths, but evidence on later-life integration is thinner. Bazzi et al. (2019) and Chyn et al. (2023) show that government resettlement programs increased minority children's later residential integration and intermarriage in Indonesia and the Chicago suburbs. Billings et al. (2021) and Kaplan et al. (2023) link school racial diversity to adult party registration.
Research on military service is closer to the present question. Green and Hyman-Metzger (2024) detect positive but statistically noisy effects of the draft on racial attitudes among about 500 white survey respondents. Ang and Chinoy (2023, 2025) find that conscription into the segregated military of World War I polarized political attitudes by race. Fryer (2007) documents a correlation between veteran status and interracial marriage but cannot separate selection from treatment. Angrist (1990) and Angrist and Chen (2011) established the draft lottery as a research design.
Data
The primary source is a 2022 national voter registration file from the consumer data firm L2. It covers about 3.6 million men born in 1950–1952, roughly 81 percent of the 4.5 million male citizens in those cohorts. About 7 percent are identified as Black or Native American and 73 percent as white, yielding analysis samples of roughly 270,000 minority men and 2.6 million white men. Hispanic voters are excluded from the minority group because about 43 percent of Hispanic citizens in these cohorts are naturalized and were unlikely to have been draft-eligible.
Several key measures are constructed rather than directly observed. A spouse is defined as a voter's only coresident at the same address within 15 years of age; about 65 percent of men have one. L2 imputes race from names and, for some Black voters, Census Block composition, using reported race in six Southern states to build the algorithm. Neighborhood integration is the 2020 minority share of a voter's Census Block. Economic outcomes include imputed income and wealth, home value, and home size, supplemented by Internal Revenue Service ZIP code income. Census and American Community Survey data supply a long-run series on interracial marriage.
Methods
The draft lottery assigned each birth date a number, and men at or below annual cutoffs of 195, 125, and 95 for the 1950, 1951, and 1952 cohorts were called up. The study compares men with and without draft-eligible birth dates, using year-by-month and month-by-day fixed effects to address the imperfect randomization of the first lottery. Because many eligible men received deferments and some ineligible men enlisted, eligibility serves as an instrumental variable for service. Prior Census estimates indicate that eligibility raised veteran status by 9.0 percentage points for non-white men and 14.5 points for white men. The resulting estimates apply to men whose service depended on the lottery.
The key assumption is that lottery status affected outcomes only through conscription. It could fail if men changed their college or marriage decisions to secure deferments, or if draft anxiety had lasting effects, though earlier studies suggest these channels are small. The study finds no relationship between eligibility and the number or racial composition of registered voters by birth date, which reduces concern about selective survival or registration. Placebo tests use men born in 1953–1955, who were not drafted, and women. Remaining weaknesses include imputed race, inferred spouses, a single 2022 snapshot, and a first-stage estimate drawn from a different dataset and a broader non-white population.
Findings/Size Effects
Among minority men, conscription raised the likelihood of having a white spouse by 3.58 percentage points from a baseline of 8.2 percent, an increase of about 44 percent, significant at the 5% level (p = 0.03). This is about 36 additional interracial marriages per 1,000 conscripted married men, although the 95 percent confidence interval is wide (plus or minus about 3.3 points). The minority share of veterans' Census Blocks fell by 1.97 percentage points from a 55.0 percent baseline, about 3.6 percent, significant only at the 10% level (p = 0.09). Republican registration rose by 2.92 percentage points from a 7.4 percent baseline, about 39 percent, significant at the 1% level, or about 29 more Republicans per 1,000 conscripted men. Democratic registration fell by 1.52 points from 85.6 percent, which is not statistically significant.
Reduced-form effects of draft eligibility were larger among Southern residents (0.52 versus 0.34 points for a white spouse, and 0.57 versus 0.27 for Republican registration), but the differences are not statistically significant and may reflect higher rates of service. In the six Southern states with self-reported race, the white-spouse effect is similar (0.45 points), while neighborhood and party effects are similar in magnitude but not distinguishable from zero. Placebo estimates are generally null. One exception is that women with draft-eligible birth dates were 0.3 points more likely to be registered Democrats, possibly reflecting sampling error.
Conscripted minority veterans also had about 9 percent higher imputed income in their early 70s (significant at the 1% level), along with about 8 percent higher home values and 8 percent more home square footage. Effects on imputed wealth, ZIP code income, and homeownership are not statistically distinguishable from zero. Under strong assumptions, economic standing accounts for only 10–30 percent of the integration and political effects.
For white veterans, effects are close to zero and precise: 0.06 points for a minority spouse from a 0.7 percent baseline, and 0.11 points for Republican registration from 48.4 percent, neither statistically significant. Neighborhood estimates rule out increases larger than 0.02 standard deviations. A simple contact model predicts effects about four times larger for minority veterans; the marriage (p = 0.06) and neighborhood (p = 0.09) results suggest even greater asymmetry, while political results do not.
Census data show Black veterans and non-veterans diverging in interracial marriage beginning with the late-1920s cohorts, the first to serve after the military integrated, with no parallel divergence for white men. If conscription's effect resembles the average effect of service, about 20 percent of all Black-white intermarriages in the early 1950s cohorts, and 25 percent of those involving Black men, resulted from integrated military service.
Conclusion
Vietnam-era conscription made Black and Native American veterans more likely to have white spouses and to register as Republicans, closing each gap with white men by 4–8 percent, and appears to have moved them into somewhat more integrated neighborhoods. It had no measurable effect on the integration or political identity of white veterans. The marriage and party results are the strongest; the neighborhood result is more tentative.
The findings do not establish that interracial contact itself was the mechanism. Conscription also raised economic standing, increased public-sector employment, and may have uprooted men from segregated home communities. The estimates apply to men whose service depended on the lottery and who were alive and registered to vote in 2022. They may not generalize to volunteer forces, civilian national service, women, or settings outside wartime. Future research would benefit from individual service records, direct measures of unit composition, separate first-stage estimates by region and race, and causal evidence on the economic channel.
Comments