Minorities have historically been marginalized in discussions about crime. In political and media debates, immigrants are often used as convenient scapegoats for complex social problems. This can lead to unfair stereotypes and policies that fail to address the real drivers of crime.
Rather than relying on rhetoric, I decided to look at the data myself. Using publicly available datasets from sources such as Eurostat, I examine whether there is any measurable relationship between immigration and crime rates.
Why Europe? Because European countries display highly diverse immigration patterns while offering relatively good data availability and methodological transparency.
Does higher immigration lead to higher crime rates?h2
Methodology
Disclaimer: The Pearson correlation coefficient is used to measure linear correlation between immigration and crime rates.
Correlation does not imply causation. Crime rates are influenced by many factors, including economic conditions, demographic change, social policy, and law enforcement practices. This analysis is purely correlational and should not be used to draw causal conclusions.
Even a strong correlation may be coincidental, and even if it is not, it does not necessarily reflect crime committed by immigrants. It may just as well reflect crime against them, changes in reporting behavior, or unrelated structural trends.
For the first analysis, I’ve pooled correlation across countries and years. Results are sensitive to structural breaks and institutional changes. Pearson r:
- is driven by means and variances
- gives disproportionate weight to extreme or sudden shifts
- assumes a single, stable relationship across all observations
A single discontinuity can:
- flip the sign
- inflate or suppress the magnitude
- create correlation where none exists
- erase correlation that exists locally
Crime type: Total
Looking at aggregate data for 35 European countries between 2014 and 2023, there is no clear correlation between immigration and crime rates, regardless of crime type. At this level of aggregation, immigration does not appear to be systematically associated with higher or lower crime.
However, countries differ substantially in their institutional frameworks, demographic structures, and migration histories. For this reason, it makes sense to examine individual countries separately.
Pearson r is very sensitive at small sample sizes. One structural break can dominate the entire result.
Austria | Crime type: Total
This breakdown shows that the relationship between immigration and crime varies widely across countries. Some exhibit positive correlations, others negative ones, and others no discernible relationship at all. The examples below are illustrative rather than definitive. Given the limited number of observations, all results should be interpreted with caution.
Portugal: r = -0.74 (strong negative)h3
As immigration increased, recorded crime tended to decrease.
This does not mean that immigration reduces crime. It means that, over time, these two series moved in opposite directions in a fairly consistent way.
Plausible explanations include:
- Crime declining due to unrelated structural factors (e.g. long-term social trends),
- Immigration increasing during periods of economic recovery,
- Demographic composition effects: immigrants are, on average, younger and often have different crime profiles than the native population. This can affect aggregate crime statistics without implying causation.
- Changes in policing practices or crime reporting.
Germany: r = 0.71 (strong positive)h3
As immigration increased, recorded crime also tended to increase.
This does not mean that immigrants caused crime, nor does it say anything about per-capita offending rates among immigrants.
Germany experienced several factors that could mechanically affect crime statistics:
- Large and sudden migration inflows
- Changes in reporting rules1
- Legalization of cannabis2
- Crime classification issues3
All of these can move recorded crime figures without reflecting a corresponding change in underlying criminal behavior.
Poland: r = 0.24 (weak positive)h3
There is a weak tendency for crime and immigration to move in the same direction, but the relationship is noisy and unstable.
In practical terms:
- The correlation could disappear if a single year is removed
- It could flip sign with minor data revisions
- It is not robust enough to support any substantive conclusions
How do non-locals contribute to crime?h2
Another way to examine the relationship between immigration and crime is to look at the share of crimes committed by non-locals, defined here as residents without local citizenship. While some immigrants eventually acquire citizenship, the majority of recent or temporary migrants remain foreign nationals, making citizenship a reasonable proxy for migration status in this context.
Citizenship is an imperfect proxy for migration status: some locals are foreign-born, while some non-locals are not immigrants in any meaningful sense. In countries with large tourist inflows, non-resident visitors may also appear in crime statistics, further blurring the distinction. These data are useful for identifying broad patterns, but they should not be treated as definitive evidence about immigrant crime.
The data show that non-locals account for a substantially smaller share of recorded crime than locals, which is consistent with their smaller population share. In countries with higher levels of immigration, the proportion of crimes attributed to non-locals is correspondingly higher, a largely mechanical effect of population composition. Even so, non-locals represent a minority of total crime across all observed countries, including those with high immigration levels.
It is also important to note that certain offenses, such as illegal entry or residence, can only be committed by non-citizens. Including these in crime statistics naturally skews the numbers. These offenses are sometimes highlighted by anti-immigrant media to create a misleading picture of “immigrant crime”4.
What next?h2
In a future update, I will examine year-over-year changes—linking Δ crimes per 100k to Δ immigrants per 1k—to see whether shifts in migration coincide with changes in crime. I also plan to explore US data, a country with historically high immigration levels, which I expect may offer more datapoints across decades. The US may also provide a useful comparison because, despite regional differences, it is relatively more homogeneous than the EU or any single European country, which have experienced numerous structural shocks in recent decades. Finally, I aim to factor in local conditions such as economic and social factors—including prejudice or xenophobia—that might influence crime rates among immigrants.