Yes, Somali Immigrants Commit More Crime Than Natives
Widespread statistics use a misleading comparison; apples-to-apples shows a two- to fivefold difference.
Since the news of Minnesota’s sprawling Somali-linked fraud cases went national, debate over immigrant crime has flared once again. The president has dispatched federal agents to the Twin Cities to crack down on illegal immigrants. But Trump is overreacting, critics contend. The Somali immigrant population, they argue, does not have particularly high crime rates.
Alex Nowrasteh of the Cato Institute, for instance, set off considerable debate on X by posting a chart showing Somali-born immigrants have, if anything, slightly lower incarceration rates than native-born Americans. Among those 18-54 captured by the 2023 American Community Survey (ACS), 1,170 of every 100,000 people born in Somalia were incarcerated versus 1,221 for the native-born.
The implication is clear. If Somalis are incarcerated at similar or lower rates, concerns about Somali crime must be overblown.
But we don’t buy this argument. Nowrasteh, we think, is not making an apples-to-apples comparison. Looking at incarceration rates introduces statistical bias in a way that yields a lower-than-expected rate of Somali offending. Correcting for this, we estimate Somalis are twice as likely to be incarcerated as are actually similar native-born Americans.
Nowrasteh’s conclusion is starkly at odds with international evidence on Somali immigrant crime rates. In countries such as Denmark and Norway, which have more thorough record-keeping than the United States does, Somali immigrants are convicted or formally charged at several multiples of native rates. If the U.S. truly had crime rates near parity, it would represent an extraordinary and unexplained divergence. What’s in the water in Minneapolis?
There is no published data for the U.S. explicitly measuring crime rates by nationality or birth country. The nation’s major crime datasets lack detailed information about immigration status. Instead, the data that Nowrasteh and others often cite on related questions come from the ACS, a general-purpose survey run by the Census Bureau of about 3 million people in the U.S. each year.
The public ACS data report whether someone is living in “institutional group quarters,” which includes prisons but also other types of institutions such as mental-health facilities and nursing homes. This isn’t a perfect measure of incarceration, but for males aged 18-40 it is a very strong proxy.
Critically, however, incarceration and crime rates are not the same. Crime rates capture the rate at which an event occurs—they are “flow” variables. Incarceration rates, by contrast, are a count of a population at any one time—they are “stocks.”
Moreover, using unadjusted differences in incarceration rates between immigrants and natives to infer their relative crime rates is not an apples-to-apples comparison. In fact, it can be deeply misleading.
Why? Consider a simple example: two groups of 40-year-old males, one American born, the other immigrants who arrived in the U.S. at age 39. The groups are otherwise equal and have the same crime rates.
Will they be equally likely to be in jail at age 40? Obviously not. The immigrants who arrived at age 39 will only have had one year to commit a crime and land themselves behind bars at age 40. The American-born, meanwhile, have had decades of opportunities to do the same.
Accordingly, in this toy example, the native-born Americans will be mechanically more likely to be incarcerated at age 40, even though the two groups have equal crime rates by design. By the same logic, because of their very different tenure in the U.S., you cannot correctly deduce—from this incarceration rate—that the immigrant group has a lower crime rate. And even if an immigrant group commits crime at very high rates,1 differences in tenure alone can still produce lower incarceration rates for immigrant groups than for native-born Americans who offend less.
In Denmark or Norway, this wouldn’t be a problem—crime rates by country of origin are measured directly. But in the U.S., if we’re forced to rely on institutionalization data as a proxy for crime, we need to confront the limitations head-on.
Failing to do this yields numbers like Nowrasteh’s (which we can narrowly replicate, though the key Somali sample size is very small), but tell you almost nothing. Such results are not evidence of equal crime rates, but artifacts of an inappropriate comparison. Treating this problem as negligible, or as unavoidable and therefore ignorable, does not make it disappear.
To produce an apples-to-apples comparison, it’s important to compare people of similar ages—and especially to avoid comparing lifelong Americans to immigrants who have spent only part of their potential offending years in the United States. Simply put, the latter have had fewer opportunities to acclimate to their surroundings, join a criminal network, build up a rap sheet, or commit serious violence, and thus land themselves in prison as adults.
To do this, we follow Nowrasteh by using the ACS but also:
analyze all of the available ACS data together (back to 2006, and through the newly released 2024 data) to boost the sample size;
narrow the scope for confounds by restricting the age range to 18-29 and look at males only (for whom institutional group quarters is a better proxy for criminal involvement);
and make a closer-to-apples-to-apples comparison by comparing the American-born to the subset of Somalis who arrived in the U.S. when they were no older than 15 (few adults are incarcerated for crimes committed before this age). Notably, this is also a more relevant comparison for second and subsequent generations of immigrants.
The results are striking. Under this apples-to-apples comparison, young men born in Somalia have approximately twice the incarceration rate of those born in the United States (5,030 vs. 2,450 per 100,000). Further, crime and incarceration rates differ sharply by race in the United States. Comparing Somali immigrants to non-Hispanic white natives (1,280 per 100,000), the difference is almost four-fold. Analysis of older age groups also reveals large disparities.
We then expand the sample to cover ages 18-64 while maintaining the same apple-to-apples comparison in a more sophisticated statistical model that controls for year (to reflect changing incarceration rates over time), the individuals’ precise ages (to account for any remaining differences in the age distribution between Somalis and natives), and state of residence (to account for differences in the punitiveness of different states’ justice systems). In this specification, the odds of a Somali immigrant being incarcerated are more than two and a half times the odds for U.S.-born males. They rise to more than four and a half times compared to native non-Hispanic whites. Given that, historically, descendants of immigrants tend to get in more trouble than the newcomers did, this is not an encouraging sign for the future.
None of this should be that surprising. Even putting aside the data from Europe, a large body of research shows that migrants do not instantly shed the behavioral and cultural norms of their countries of origin. Raymond Fisman and Edward Miguel famously showed this in a study measuring unpaid parking tickets accrued by U.N. diplomats in New York: officials from more corrupt countries behaved far more corruptly, even under identical enforcement conditions, and these differences persisted over time.
Alberto Alesina and Paulo Giuliano, writing in the Journal of Economic Literature, concluded that “when immigrants move to a place with different institutions, overwhelmingly their cultural values change gradually, if ever, but rarely within two generations.” Transparency International, in its Corruption Perceptions Index, ranks Somalia 179th (out of 180) in the world. Simply put, there is a large institutional and cultural gap between Mogadishu and Minneapolis.
Thanks to the limits of the ACS, we readily concede that our statistical analysis is still limited and fails to estimate a “Somali crime rate” with any precision. Measurements of crime by birthplace or immigrant status would be better served by statistics that do precisely that, but this may be the only way to credibly measure crime among those who arrived as adults.
What is clear, though, is that the data do not justify dismissing public concerns as innumerate fearmongering. Rather, they clearly show that, under an apples-for-apples comparison looking at those with comparable time spent in the U.S., Somali immigrants have incarceration rates far above the U.S. native-born average.
For example, as The Princess Bride reminds us, “as everyone knows, … Australia is entirely peopled with criminals.”
Taking a snapshot of incarceration has another flaw: It’s affected by “restorative justice” judges handing out shorter sentences to their client groups.
"In countries such as Denmark and Norway, which have more thorough record-keeping than the United States does, Somali immigrants are convicted or formally charged at several multiples of native rates. If the U.S. truly had crime rates near parity, it would represent an extraordinary and unexplained divergence. What’s in the water in Minneapolis?"
Aren't native born Americans much more likely to commit crimes than native born Danes, or Norweigans? I'd have guessed that if Somali immigrants committed crime at the same rates regardless of where they immigrated to, you'd still see big differences in the ratios of Somali crime rates to native born rates based on differences in crime in the places they're immigrating to. Am I missing something?