The racial integration of the US Army during the Korean War (1950-1953) is one of the largest and swiftest desegregation episodes in American history. This paper argues that racial integration improved white survival rates at the expense of blacks, and resulted in less anti-black prejudice among white veterans decades after the war. Using a novel military casualty file, I construct a wartime similarity index to measure the extent of racial integration across military units and time. Using exogenous changes in racial integration, I show that integrated whites were 3% more likely to survive their injuries than segregated whites, whereas integrated blacks were 2% were less likely to survive their injuries than segregated blacks. Given that blacks were initially confined to noncombat support roles, the results reflect a convergence in hazardous combat assignments. To explore the long-term effects of racial integration, I link individual soldiers to post-war social security and cemetery data using an unsupervised learning algorithm. With these matched samples, I show that a standard deviation change in the wartime racial integration caused white veterans to live in more racially diverse neighborhoods and marry non-white spouses. In aggregate, these results are some of the first and only examples of large-scale interracial contact reducing prejudice on a long-term basis.
That is from the job market paper from Daniel Indacochea of the University of Toronto.