This paper analyzes the impact of immigration on robbery and burglary rates while differentiating between European Union (EU) countries with relatively large and small immigration rates using a dynamic panel data model. Notably, robbery (i.e., “stealing from someone by using physical force, weapon or threat”) differs from burglary (i.e., “breaking, entering and stealing from homes, summerhouses, hotel rooms, cabins”) in its use of force (Eurostat, 2022, https://ec.europa.eu/eurostat/statistics-explained).

Several crime studies considered immigration, but only as an additional covariate in their models. They found mixed outcomes of positive correlations (Alonso-Borrego et al., American Law and Economics Review, 2012; Solivetti, Quality and Quantity, 2018), negative correlations or no associations (Ozden et al., The World Bank Economic Review, 2019; Leiva et al., World Development, 2020). Papadopoulos (IZA Journal of Migration, 2014, p. 6) noted that while having few employment opportunities may motivate immigrants to commit property crimes, which he defined as “thefts and attempted vehicle thefts, thefts and attempted thefts of parts from inside or outside vehicles, robberies, burglaries, thefts [from] person, work, school, shops, other thefts and criminal damage”, undocumented immigrants living under the threat of deportation may be more law-abiding relative to natives. Cohn et al. (Italian Review of Criminology, 2021) investigated the relationship between socioeconomic factors and violent and property crimes for 39 European countries over three years, separating nations based on high and low immigration rates. However, they only discussed descriptive statistics with respect to the two groups.

The current sample includes annual data from 1993 to 2019 for 20 EU countries based on data availability. The nations are sorted by immigration subgroup (Online Supplemental Appendix (OSA) Table 1). The dependent variables are per capita rates of robbery and burglary (Eurostat, 2022, https://ec.europa.eu/eurostat). The key regressor is the per capita legal immigration rate (lnimm) (World Bank, 2022, https://data.worldbank.org/indicator/SM.POP.TOTL). Other covariates gathered from the World Bank Indicators (World Bank, 2022, https://data.worldbank.org/indicator) included the unemployment rate (lnun), per capita gross domestic product (GDP) (lngdp), the urbanization rate (lnurban), and the Gini coefficient (lngini). Models also incorporated police officers per capita (lncop) (Eurostat, 2022, https://ec.europa.eu/eurostat), a euro dummy (euro), and a lagged dependent variable (lnrob(-1) or lnburg(-1)) measuring criminal inertia. All non-dummy regressors are in natural logarithms with coefficients interpreted as elasticities (Butkus et al., Proceedings of the Faculty of Economics in Rijeka: Magazine for Economic Theory and Practice, 2019).

To account for endogeneity, measurement error, heteroskedasticity, and serial correlation, generalized method of moments (GMM) estimation was used to control for unobserved heterogeneity and to obtain efficient estimates (Arellano & Bond, Review of Economic Studies, 1991). A Chow test rejected the null hypothesis of equal coefficients in the high immigration and low immigration regressions for robbery and burglary (p < 0.01). Models included the full sample and high and low immigration subsamples for each dependent variable.

According to the results (OSA Tables 2 and 3), immigration is not significant in any of the robbery models. This result may not be surprising. Many studies find no relationship between immigration and violent crimes, and robbery involves physical force or the threat of force. In contrast, with the full sample, a 1% increase in the per capita immigration rate increases per capita burglaries by 0.07% (p < 0.1). More notably, while no significant relationship exists in the low immigration subgroup, a 1% increase in immigration in the high immigration subgroup increases per capita burglaries by 0.27% (p < 0.01). This positive relationship in countries with high immigration rates may reflect social disorganization theory (Han & Piquero, Crime & Delinquency, 2022), which asserts that differences in culture, tradition, or language may limit interactions between native citizens and immigrants (Charis & Ronald, Recent Developments in Criminological Theory, 2017). This reduced community connectiveness may increase the likelihood of immigrant criminality. Alternatively, immigrants may experience greater difficulty securing employment relative to locals (Reid et al., Social Science Research, 2005), and these employment rate differences may exacerbate the number of financially motivated crimes committed by immigrants.

Each lagged dependent variable carries a positive and significant coefficient (p < 0.01) across all models. In low immigration nations, both robbery and burglary rates decline with unemployment (p < 0.05 and 0.1, respectively) and GDP (p < 0.1 and 0.05, respectively) while robbery rises with GDP (p < 0.01) in high immigration countries. Interestingly, police officers (p < 0.05) seem to deter burglary in low-immigration nations only.

These findings may inform policy decisions for countries contemplating joining the EU. However, appropriate crime-fighting measures may be sensitive to the degree of immigration experienced by a country. Thus, differentiating samples based on low versus high immigration rates may offer a more informative approach to explaining certain crimes in the EU.