The relationship between social order and crime in Nottingham, England

The paper examines whether a governance-type organized crime group (OCG) can govern a local community by reducing ordinary crime in a context with a strong state and a nonimmigrant, less affluent population. Prior research on criminal governance has focused largely on Latin America, traditional mafia territories, or immigrant enclaves, often assuming weak state capacity or lack of access to legitimate institutions. The authors study Nottingham, England, to test these assumptions, focusing on Bestwood, where interviews and documentary evidence indicate the presence of a long-standing governance-type OCG linked to Colin Gunn. The study uses a novel dataset of public calls to police (2012–2019) categorized into eight crime types (CTs) and combines macro-level spatial analysis with qualitative interviews. The central research question is whether the presence of a governance-type OCG is associated with reduced levels of ordinary crimes compared to a demographically and socioeconomically similar ward without such an OCG (Bulwell). The work aims to refine theories of criminal governance by showing it can emerge and exert community-level effects within the global north under conditions of high formal policing capacity.

The study builds on a long tradition of research on organized crime and criminal governance, from early work by Landesco on Chicago to studies of traditional mafias (e.g., Sicily, Russia, Japan) and urban gangs in the United States, and more recent analyses of Latin American cartels and combos. This literature documents how OCGs sometimes provide governance by imposing social order, resolving disputes, and protecting business activities, coexisting with state institutions. Many studies emphasize contexts with weak state infrastructure or marginalized immigrant communities. Recent UK work has begun to examine criminal governance domestically. The authors highlight that existing explanations—emphasizing state weakness or immigrant isolation—may not fully account for the emergence of criminal governance in places like Bestwood, a relatively poor but nonimmigrant ward in an English city with high policing capacity. Their mixed-methods approach responds to calls for combining ethnographic insight with large-scale data to map local variations in criminal governance and its effects.

Design: Mixed-methods study combining quantitative analysis of police call-for-service data with qualitative fieldwork and documentary review. Ethical approval obtained from the University of Oxford (R70851/RE004); informed consent from all interviewees. Data sources: (1) Anonymous dataset of public phone calls to Nottingham police, 2012–2019, including timestamp, postcode location, and police-labeled crime types (CTs). Data aggregated from postcode to Lower-layer Super Output Areas (LSOAs) and then to 20 Nottingham wards using pre-2019 ward boundaries. (2) Socioeconomic data from the 2011 UK Census and the 2015 English Indices of Deprivation, aggregated from LSOA to ward. (3) Thirteen interviews (2022–2023) with senior police officers regarding OCG presence and operations; field visits to observe the built environment; analysis of books and news sources. Crime types: Eight CTs selected based on protection theory and visibility/reporting: CT1 antisocial behavior; CT2 burglary/robbery/theft; CT3 civil dispute; CT4 criminal damage; CT5 drugs; CT6 violence against the person; CT7 firearms; CT8 knives/bladed weapons. Analytical steps: (a) Descriptive spatial analysis of average crime counts (2012–2019) by ward; assessment of temporal stability at LSOA level. (b) Multiple correspondence analysis (MCA) to explore associations between wards and CTs; biplot uses the two principal dimensions capturing 82.9% of inertia. Implemented with Python Prince 0.13.0. (c) Matching comparison: Identify a ward similar to Bestwood on nine socioeconomic features (population density; % aged 15–24; % with Level 4+ qualifications; % born outside UK; % social rented; and IMD domains for income, employment, health/disability, and barriers to housing/services). Features scaled by per-feature maxima; similarity measured with Minkowski metric (p=1 Manhattan; p=2 Euclidean) using scikit-learn 1.3.2 k-NN. Bulwell is nearest neighbor under both metrics. (d) Statistical inference: Compare Bestwood vs Bulwell on crime counts per population density using two-sided Welch’s t-tests with 10,000 bootstraps per CT. Additional analyses include chi-squared tests on average crime counts, and comparisons to Bilborough (second-nearest neighbor) and adjacent wards (Basford, Bulwell Forest, Sherwood). (e) Qualitative synthesis: Interviews and field observations to characterize OCG governance in Bestwood and to assess built environment similarities with Bulwell. Software: Python 3.8.8 with csv, geopandas, matplotlib, numpy, pandas, Prince, seaborn, scipy, and scikit-learn. Sample bootstrap code provided in Supplementary Code.

• Citywide patterns: Central wards (Bridge, St Ann’s) show high levels of all CTs; western suburban wards (Wollaton West, Wollaton East and Lenton Abbey) show low crime; peripheral north shows intermediate crime with variation (e.g., Leen Valley higher firearms; Bulwell higher criminal damage than Bestwood). Overall LSOA crime trends were relatively stable 2012–2019.
• Governance-type OCG in Bestwood: Interviews and documentary sources identify a long-lived, governance-type OCG led historically by Colin Gunn, exerting informal control, resolving disputes, policing ordinary crimes, and engaging in intimidation (often via criminal damage). Despite the 2006 arrest of Gunn, interviewees report continued influence from prison and enduring reputation effects.
• MCA: The two principal dimensions capture 82.9% of inertia, mapping associations between wards and CTs and indicating clusters of similar crime profiles; suburban wards appear distant from most CTs (low crime).
• Matched comparison (Bestwood vs Bulwell): Using nine socioeconomic features, Bulwell is the nearest ward to Bestwood. Bestwood exhibits significantly lower crime counts per population density than Bulwell across all eight CTs over 2012–2019 (two-sided Welch’s t-tests with 10,000 bootstraps; P<0.05 or P<0.01 as shown in Fig. 3). Reductions are notable for CT1 antisocial behavior, CT2 burglary/robbery/theft, CT3 civil dispute, CT4 criminal damage, and CT6 violence against the person. CT5 drugs, CT7 firearms, and CT8 knives/bladed weapons are lower in Bestwood as well, though volumes are low and differences are smaller.
• Temporal persistence: Yearly comparisons indicate Bestwood remains below Bulwell throughout 2012–2019, suggesting the gang’s capacity to police certain crimes persisted after the leader’s arrest.
• Robustness: Chi-squared tests on average counts show significant differences between Bestwood and Bulwell across CTs; comparisons with Bilborough and neighboring wards generally confirm lower rates in Bestwood for key CTs.
• Built environment: Field visits and public touring videos indicate Bestwood and Bulwell have comparable built environments, suggesting physical differences are unlikely to explain crime disparities.

The study’s central question—whether a governance-type OCG can govern by reducing ordinary crime in a high-capacity state context—is answered affirmatively in the Nottingham case. Bestwood, identified qualitatively as under the influence of a governance-type OCG, shows significantly lower rates of multiple ordinary crimes relative to a closely matched ward (Bulwell). These effects are strongest for expressive and business-related offenses (antisocial behavior, burglary/robbery/theft, civil disputes, criminal damage, and violence against the person), consistent with protection theory: OCGs enforce behavioral norms, deter local predation, and protect business activities to maintain order and legitimacy. Crimes more directly tied to illicit markets or tools (drugs, firearms, knives) are lower but closer in magnitude, reflecting that revenue-generating activities are not necessarily suppressed. The persistence of lower crime rates after the leader’s imprisonment suggests durable reputational and network effects. Citywide, the mapping and MCA reveal a heterogeneous urban crime ecology: central commercial areas draw higher crime, suburbs remain low, and the north shows varied patterns, underscoring that OCG impacts are highly localized. The findings challenge dominant explanations that tie criminal governance to weak states or immigrant enclaves, demonstrating its emergence in a nonimmigrant, less affluent area within an English city. Considerations of reporting behavior and anonymous hotlines suggest that the observed differences are unlikely to be solely due to suppressed reporting. The study emphasizes the importance of incorporating both qualitative institutional knowledge and large-scale data to detect and interpret criminal governance in advanced urban settings.

This study documents a governance-type organized crime group operating in Bestwood, Nottingham, and shows through matched quantitative comparisons that ordinary crime rates are significantly lower there than in a socioeconomically similar ward without such an OCG. By combining police call data, correspondence analysis, supervised matching, statistical testing, and in-depth interviews, the paper extends the criminal governance literature to a global-north, nonimmigrant context with strong state policing capacity. The principal contribution is to challenge traditional explanations of criminal governance and to highlight that such governance can coexist with robust state institutions while shaping local crime patterns. Future research should pursue causal identification strategies (e.g., natural experiments or panel methods with exogenous shocks), examine mechanisms of compliance (fear, legitimacy, service provision), assess generalizability across UK and other global-north cities, and evaluate the effects of targeted community policing and social service expansions (e.g., Operation Reacher) on displacing or dismantling criminal governance.

• Measurement: Outcomes are based on public calls to police, which may diverge from actual crime occurrences and official recorded crime. Selection of CTs focused on visible/reportable events; domestic or hidden crimes are not captured.
• Reporting behavior: While anonymous reporting for drugs, firearms, and knives appears similar across areas, unobserved differences in reporting propensity could bias comparisons. The authors argue large fear-induced reporting suppression is unlikely, but it cannot be fully ruled out.
• Observational design: Causality cannot be definitively established; matched comparisons mitigate but do not eliminate confounding and reverse causality concerns.
• Geographic units: Analyses use pre-2019 ward boundaries; later boundary changes likely have minimal impact but could introduce minor misalignment.
• Generalizability: Findings are specific to Nottingham’s context, particularly Bestwood’s long-standing OCG and local institutions; extrapolation requires caution.
• Data access: The call-for-service dataset is not publicly shareable, limiting external replication; summary code and public covariates are available.