Thread by @cremieuxrecueil

Between 1980 and 2020, there were nearly three-times as many Black-on-White as White-on-Black homicides.

But there are ~5x as many Whites as Blacks, so doesn't that mean Blacks are at ~1.92 greater risk of interracial violence? No, because of segregation.

Blacks and Whites tend to live in different areas. If you’re familiar with the concepts of “White Flight” or “Redlining”, this shouldn’t be a shock to you. If you’re not, then think about what happened with education in the last century.

As Welch & Light recorded in 1987, White school enrollment declined when desegregation efforts started. Sarah Reber recorded that about one-third of the positive effect of desegregation plans was offset by declines in White enrollment shares. Bryan Lutz found that after court-ordered desegregation programs ended, racial segregation trended up, partly due to increasing Black dropout rates and private school attendance. Reardon et al. replicated these results for segregation by showing that it increased after court ordered desegregation ended, relative to districts where desegregation orders remained in place.

Desegregation programs certainly reduced segregation in schools, but they also contributed to segregation between districts because people moved in order to maintain segregation. This is not the only reason White Flight happened, but it is a sufficient example to show that people do things that lead them to live in segregated areas.

Thanks to segregation, the numerator is not five and the denominator is not 2.6. The county population-weighted White:Black ratio for population across all counties is 2.17, and the same ratio for homicide victimization is 2.70. This means that risk is reversed once we take segregation into account.

Light & Thomas used five decades of data on segregation and homicide rates across 103 metropolitan areas to assess whether Whites benefit from segregation through reduced crime rates. Their conclusion was:

"As our analysis demonstrates, the black-white homicide gap in the least segregated metro areas is a fraction of the gap in the most segregated areas... decreased segregation over [the] period [1970-2010] reduced the [black-white] homicide gap by over 25 percent, net of controls for black-white differences in disadvantage, affluence, and all other controls from our main analysis, including MSA and year effects."

In other words, the more Blacks in the area, the higher the homicide rates for Blacks *and* Whites. And as desegregation has taken place, the homicide rates experienced by Whites have increased, while the ones for Blacks have declined.

We can replicate this straightforwardly using county-level data. Notice how, the higher the Black share of the county, the smaller the Black-White homicide rate gap.

You might say 'Ah, well, those places with more Blacks are poorer, so of course those Whites will be killed/kill more often.' It's not that. Control for income, and the result holds:

You are nearly twice as likely to be murdered in a county that's 40% Black than in one that's <5% Black.

Results in this direction are entirely predictable if you're familiar with basically any stats on homicide in the U.S. (x.com/cremieuxrecuei…) You have to be daft to miss that race dominates in explaining American homicides. But, luckily, we have a simple solution to make it so race no longer dominates so decisively, reducing homicide rates greatly in the process:

Incarceration and policing.

You only need to lock up a few percent of the population to dramatically reduce homicide rates. It's going to be a higher share of the Black population in the U.S. and a lower share of the Asian population, relative to the White one, but that's just how it is. The sooner we accept that, the sooner we can render race irrelevant, reclaim urban areas from crime (and the perception of crime), and bring America towards more reasonable homicide rates.

Source: cremieux.xyz/p/serial-kille…

Further reading: x.com/cremieuxrecuei…