Immigration and Crime in Europe

When I wrote a post about immigration and crime in the US, I promised a followup with other countries. A mere four months later, here it is.

An important caveat to start: the evidence for each individual country is much scarcer here. The US has 350 million people and 16,000 economists;¹ the UK has fewer than 70 million people and maybe a couple thousand economists. This scales accordingly with other countries; my conclusions for each country are therefore much weaker. Still, I’ll make a stab at summarizing what we do know.

Several caveats upfront: I only speak and read English. This restricts the amount of information I have on some countries. I also focus here, as I have in other posts, on the academic causal evidence. Much of the reporting on immigration and crime examines the correlation between immigration and crime. That is interesting enough, but I want to know if immigration causes crime.

The UK

In my last post, I included this chart:

(from Marie and Pinotti 2024, though the FT also has a variant on this chart)

In this chart, you can see an Anglosphere cluster; in the UK, the USA, Australia, and New Zealand, migrants are underrepresented in prisons and jails.

As I’ve written before, this does not conclusively prove that immigrants in the UK commit less crime than natives. If foreign criminals are deported rather than being incarcerated in the UK, this could lead to fewer criminal immigrants in prison than criminal natives.

It seems like deportation does happen in at least some cases; about 3,500 foreign national criminals are deported from the UK each year. However, I don’t think this is incredibly significant, given the following:

• The foreign population in prison has been relatively constant from 2014 to 2024, suggesting that each year, the deported population is (roughly) replaced with new prisoners. So we have ~3,500 new prisoners each year.

• The average prison sentence length is 20 months, half of which is served in prison. So let’s say the average prisoner is in prison for about a year.

• Thus, if no prisoners were deported, you’d have about 3,500 more people in prison that year - and the years following. This would increase the foreign criminal population to ~14,000.

• There are about 100,000 prisoners in the UK, so in a case where there were no deportations, the prison population would be 14% foreigners.

• The UK is 16% foreign-born.

So even if there were no deportations at all, migrants would still be underrepresented in prisons and jails.

If migrants were less likely to be sentenced to prison time than the native born, this might also lead to lower imprisonment rates than would be warranted by crime rates. I think this is quite unlikely in the UK context. Ethnic and racial minorities are more likely to be stopped by police and more likely accused of crimes than white people. Not all migrants are racial or ethnic minorities,² and not all racial and ethnic minorities are migrants, but the categories do often overlap.

I haven’t been able to find good studies about bias in the criminal justice system relating to migrants in particular, but it seems more plausible to me that migrants would face bias against them than towards them in the UK criminal justice system.³ Thus, I think the descriptive statistics suggest it is reasonably unlikely that migrants commit more crimes than the native born in the UK.

The causal evidence is slightly less optimistic than this; it doesn’t particularly show that migrants commit less crime than the native-born in the UK, but nor does it show that migrants commit more crime than the native-born. Rather, it seems that a significant rise in migration over the 1990s and 2000s had basically no impact on crime in the UK.

One paper looks at two different immigrant waves - the late 90s/early 00s wave of refugees from Iraq, Afghanistan and Somalia, and a wave of immigrant from countries that recently joined the EU in 2004. They find a small increase in property crime from the first wave, and a small decrease in property crime from the second wave. They find no change in the rate of violent crime from either.

Another paper also looks at migration in the mid-to-late 00s. It finds no impact of migration on crime - even if one considers only EU migrants, only non-EU migrants, or only London.

It is true that migration to the UK has changed since 2004, particularly after Brexit in 2016. The academic literature doesn’t update terribly quickly, so I can’t say for certain what such designs would find now. However, since migrants are still underrepresented in prisons,⁴ I think it is likely that immigration has a null-to-negative effect on crime in the UK.

Germany

Germany is The Case Study on immigration and crime. In 2015, about 890,000 Syrians applied for asylum in Germany. For a country that usually receives a total 300,000 asylum applications, this was a massive surge in immigration.

Germany’s Syrian population is also something of a worst case scenario for crime. The majority of Syrian refugees in Germany are young men, the group that tends to be most involved in criminal activity. In 2024, the unemployment rate for Syrians was 37%; the unemployment rate for Germany as a whole was 6.4%. Lots of unemployed young men? It wouldn’t be incredibly surprising if that did lead to more crime.

Here is what we know:

• Immigrants are overrepresented in German prisons.

As I note above, this is relatively common in the non-Anglosphere. It is possible this means that migrants are more likely to commit crimes. It may also be that migrants are more likely to serve prison time than the native born for the same offense, be that because of racism in either conviction or sentencing or simple unfamiliarity with the legal system. So, this is suggestive that refugee intake may have increased crime, but not particularly conclusive.

• A relatively early analysis by Gehrsitz and Ungerer - including only 2014 and 2015 - showed an increase in fare-dodging and drug crimes, but not other types of crime.

• Huang and Kvasnicka 2019 finds that in some specifications,⁵ increases in the number of refugees caused increased crime; in some specifications, it did not.⁶

They do find that in all specifications there was no increase in German victimization; that is, if migrants cause more crime, they are committing it (largely) against other migrants, not the native born.

• Maghularia and Uebelmesser 2019 finds no link between immigration and crime over a longer time period (2003-2016). Interestingly, they find immigration increased crime in the first part of the period, but may have decreased crime during the refugee crisis.

• Lange and Sommerfeld 2020 finds that Syrian migrants did increase crime rates in the medium term, though they also find some evidence that Syrians are overreported for crimes.

• Dehos 2020 finds no impact in the crime rate from asylum seekers, but that once granted legal status as refugees, these immigrants are more likely to commit crimes than the native born. (The latter result uses a shift-share instrument, and as I have previously discussed, I don’t love shift-share instruments in migration, so that part of the paper I’m quite skeptical on.)⁷

What conclusion do I draw from this mixed evidence? Honestly, I’m not sure. It seems possible to me that resettling one million Syrians in German raised the crime rate; it also seems possible that it didn’t. Frankly, I’d need to spend more time understanding how these papers get such different results - and going through their data sets and empirical strategies - to come to much of a conclusion here.

Italy

As in Germany, foreign nationals are overrepresented in Italian prisons. However, as in Germany, this doesn’t tell us much; is this because foreigners commit more crime, is it that the criminal justice system is racist, or is it that foreigners commit immigration offenses (which by definition, the native born cannot commit)?

Bianchi, Buonanno, and Pinotti 2012 shows that the causal effect of immigration on crime across Italian provinces is not significantly different from zero. However, it is an instrumental variable paper, so it’s not the strongest evidence. A 2021 paper also shows no link between likelihood of being victimized in a crime and immigration, though they do find a link between perception of crime and immigration.

This isn’t really a section about the relative likelihood of foreigners in Italy to commit crimes, though;⁸ it’s mostly an excuse to talk about Pinotti 2012.

This paper doesn’t look at the likelihood of immigrants to commit crime writ large; rather, Pinotti 2012 looks at how circumstances can change an individual’s propensity to commit crimes. If you are a country who is worried about immigration and crime, is there anything you can do to make this less likely?

In Italy, undocumented migrants could apply for a (legal) residence permit if they submitted an application online on a particular day.⁹ Only a certain number of residence permits were available, and demand exceeded supply. Applications would begin at 8 AM; by 8:30 AM, all available residence permits would be taken and subsequent applications rejected.

Pinotti compares applicants who applied just before the quota of residence permits was exhausted and those who applied just after. Migrants wouldn’t know which category they were in until after submission; approvals continued until residence permits ran out, and it wasn’t possible to tell ex ante when that would be. It was random who became documented and who stayed undocumented.

In short, undocumented migrants that became documented were much less likely to commit crime after they became documented. The magnitude of the change was significant; the crime rate among newly legalized immigrants dropped to half that of those that did not get legal status.

Why? The usual economic model is that individuals choose between legal and illegal activity as a way of generating utility. If you’d make more and be happier as a criminal, you’ll be a criminal. If you’ll make more and be happier getting a normal job, you’ll get a normal job.

Undocumented migrants don’t have great job opportunities. They must work under the table, often in positions that are not particularly well-matched to their skills and with employers that may treat them badly.¹⁰ The returns to the non-criminal sector aren’t great. In general, undocumented migrants still prefer an under the table job to entering a life of crime - in Pinotti’s sample, at least 98.9% of undocumented migrants hadn’t committed a serious crime in the last year - but it is at least plausible being a criminal would pay more than working as a undocumented migrant.

Once regularized, migrants have access to a whole new universe of jobs. These jobs generally pay better and have better working conditions, and crime looks less good by comparison. Why take the risk that you might be thrown in jail if you can have afford to have a perfectly nice life and not be thrown in jail?

There’s some empirical evidence that this is true. Elias, Monras, and Vázquez-Grenno 2022 examines the regularization of about 600,000 migrants in Spain (equivalent in per capita terms to the US regularizing ~5 million migrants). Most migrants promptly got a legal job instead of sticking with their previous under-the-table job and started paying quite a lot more in taxes (~4000 euros each). These gainfully employed migrants would have a lot more to lose if they got arrested than those working under the table.

This suggests that there is one weird trick to keep immigrants from committing crimes. Migrants with access to good labor markets - where they can earn more money from being gainfully employed than from being a criminal - are less likely to commit crimes.

What about France and Belgium?

When discussing immigration and crime in Europe, the above countries aren’t really the center of discussion. Instead, it tends to be around particular neighborhoods in the outskirts of Paris, and Belgium. Why don’t I have sections on those countries?¹¹ Is it because I am unwilling to consider that immigrants might raise crime rates?¹²

It’s actually because I don’t have great data. There’s Todo and Aoki 2009 on France, but that uses only data from 1999. One study focusing on data from 26 years ago is not much evidence! I have even less from Belgium; I couldn’t find any good causal papers.

Conclusions

As always, the academic literature is a bit unsatisfying. There are many different immigrant communities in many different countries; there is not one answer to how immigration affects crime.

Immigration probably does not increase crime in the UK (or indeed, any other Anglosphere country, including the US). It might increase crime in Germany; I’m not sure. I don’t really have enough evidence to say either way about France, Belgium, or Italy.

It does appear to be robustly true that immigrants are more likely to commit crimes the less they have to lose. If they are working and able to make more from a job than they could from committing crime, they are considerably less likely to commit crimes. If you are a policymaker who is concerned about immigration and crime, one could do worse than making sure immigrants were gainfully employed, paying taxes into your economy and possibly a bit too busy to engage in criminal activity.

Many thanks to Akib Khan for his extremely thoughtful feedback. All wild generalizations that remain are my own responsibility.

1. Though this number excludes academic economists, who, according to academic economists, are the only real economists.

2. Hi! I’m a migrant in the UK! I am also very, very white.

3. For a start, speaking the same language as your lawyer and the judge tends to help in criminal proceedings.

4. As of 2024.

5. A flexible non-linear specification.

6. A linear specification

7. Especially since it conflicts with other evidence from Italy and Spain, discussed in the next section. The asylum seeker result uses a different identification strategy, so I’m less skeptical there.

8. For a start, I don’t like either of these papers and thus don’t feel like I can really answer that.

9. Theoretically, they needed a real job offer. In practice, fake job offers were common, so employment was not truly required.

10. After all, what is the migrant going to do if their boss just doesn’t pay them? Going to the police might not be an option.

11. If you are wondering instead about [insert country here], feel free to email me at lagilbert@gmail.com to request I cover a specific country.

12. Probably not, since I wasn’t willing to say that wasn’t the case for Germany.