A UC Irvine study estimated undocumented population shares for 11,500 US neighbourhoods from 2010 to 2018 and found property crime fell further, violent crime unchanged, and robberies up.

Whether immigration raises crime is among the most studied questions in criminology and among the least settled in public argument. The research literature has largely found no criminogenic effect and often a protective association, and almost all of it has been conducted at the level of states or metropolitan areas — which leaves open the objection that a city-wide average could conceal what happens on a particular street.

A study in the Journal of Urban Affairs from the University of California, Irvine addresses that objection directly, estimating undocumented population shares neighbourhood by neighbourhood across 11,500 neighbourhoods containing 46 million residents, from 2010 to 2018.

The lead author is Charis E. Kubrin, Chancellor’s Professor of Criminology, Law and Society at UC Irvine, with Derek Christopher of Stanford University, Cheyenne Hodgen of UC Irvine, Xiaoshuang Iris Luo of the University of Akron, and John Hipp of UC Irvine.

The hard part is the denominator

Nobody is asked their immigration status on a form that researchers can read, so estimating how many undocumented residents live in a given census tract requires inference from characteristics that are recorded.

Christopher, who adapted the method, is matter-of-fact about its provenance. “Different versions of these methods have been around for decades,” he said. “How else would anyone be able to know roughly how many undocumented people are in the United States?”

He said that “our study, to my knowledge, is the first to apply these methods to the rich data that only exists in the RDC” — the Federal Statistical Research Data Centers, which hold restricted census microdata that cannot be analysed outside secure facilities.

This is where the study’s novelty and its main vulnerability both sit. Access to that microdata is what permits neighbourhood-level estimation at all; the estimates remain statistical imputations rather than counts, and any error in them propagates into the crime analysis. The method is the established one in the field, which makes it defensible and not exact.

What was found

Neighbourhoods where the undocumented share grew saw greater reductions in property crime than neighbourhoods where it did not.

Violent crime overall showed no significant change. Within that, aggravated assaults declined significantly — and robberies increased.

“Consistent with the research base more generally, unauthorized immigration does not drive crime rates in U.S. neighborhoods,” Kubrin said. Asked about the result, she added: “Honestly, I wasn’t too surprised.”

She placed it in a long lineage. “Nearly 100 years ago, the Wickersham Commission published a report showing that involvement in crime is lower among immigrants than the native born,” she said, “a finding that made its way onto the front page of the New York Times.” The reference is accurate — the 1931 federal commission did report that — and the persistence of the pattern across a century of American research is among the more robust findings in criminology.

The robbery exception

The one finding that moved against the pattern is the one the authors spend most effort explaining, and their explanation is a claim about victims rather than offenders.

“An increase in robberies, despite decreases in all other crime categories, would be consistent with the idea that undocumented immigrants are disproportionately likely to be targets of robbery given greater involvement in the cash economy,” Hipp said.

The reasoning is coherent and has an evidence base behind it: workers paid in cash who are reluctant to use banks, and who may avoid contact with police, carry money and make attractive targets — a pattern documented in research on immigrant victimisation.

It is also an interpretation rather than a measurement. Crime data of this kind records offences by location, not the immigration status of victims or offenders. The authors’ explanation is the one most consistent with the rest of their results — a population committing more robberies while committing fewer assaults and fewer property crimes would be an odd pattern — but the data do not adjudicate it directly.

What the study can and cannot support

The design limits are worth stating clearly, because the question attracts stronger conclusions than any single study can carry.

These are neighbourhood-level associations, not individual-level findings. Attributing an area’s rates to the behaviour of particular residents is the ecological inference problem, and it can mislead in either direction. Neighbourhoods also change in many ways at once — housing costs, age structure, employment, policing intensity — and the analysis’s ability to separate those from the variable of interest depends on controls the release does not detail.

Reported crime is also not crime. If residents of a neighbourhood become less likely to call police — which is a documented consequence of intensified immigration enforcement — recorded offences can fall without anything improving. That mechanism would produce exactly the pattern observed in property crime and assaults, and it is the alternative explanation a reader should hold alongside the authors’ own.

One administrative oddity: the release’s own metadata lists the method of research as a meta-analysis, which does not match the neighbourhood-level analysis of census and crime data it describes. That is most likely a tagging error on the release rather than a statement about the paper.

Neither funding sources nor conflicts of interest are stated.

The policy claims, and where they sit

Kubrin draws conclusions beyond the findings, and they should be identified as her position rather than as results.

“It’s well beyond time to rethink immigration enforcement in this country,” she said. “These policies are largely ineffective at enhancing public safety and may actually do more harm than good.” The release pairs this with a separate study led by T. William Lester of UC Irvine, reporting that Orange County businesses lost nearly $59 million in economic output over eight weeks following intensified federal immigration enforcement. “With policies like these, no one wins,” Kubrin said.

Those are statements about what policy should be, and they rest on premises this study does not test. The paper examines whether undocumented population share predicts neighbourhood crime rates; it does not evaluate any enforcement policy, measure deterrence, or weigh the other objectives — labour market effects, fiscal costs, rule-of-law arguments, the interests of prospective migrants waiting lawfully — that people on various sides of the debate consider relevant.

Those who favour stricter enforcement generally ground the case in legal and sovereignty arguments, in wage competition for lower-paid native and lawfully resident workers, or in fiscal burden, rather than in a claim that undocumented residents commit more crime — so a finding about crime rates, however solid, leaves much of the actual disagreement untouched. Critics of research in this area also contest the status-imputation methods themselves, since the estimates cannot be validated against a ground truth that does not exist.

What the study adds is a sound and more finely resolved answer to one empirical question: at the neighbourhood scale, on these estimates, a growing undocumented share did not come with more crime. That is a useful fact, and it is not by itself a policy.

The paper was published on 10 August 2026.

---

Notes

1. Kubrin, C.E., et al. “Does unauthorized immigration drive community crime rates? Evidence from 11,500 U.S. neighborhoods.” Journal of Urban Affairs, 2026. DOI: 10.1080/07352166.2026.2699123.
2. University of California – Irvine. “Large-scale study finds no link between crime and unauthorized immigration.” EurekAlert!, 10 August 2026. https://www.eurekalert.org/news-releases/1139581.

Photo by Charles Parker on Pexels.com