Pubs and pints, crims and crimes: exploring the relationship between public houses and crime

ABSTRACT

This paper examines the relationship between public houses or pubs, and crime rates in England and Wales. The impact of pubs on local communities is generally studied and investigated within the context of third places, thus physical places that facilitate the accumulation of social capital within communities. We estimate Poisson Fixed-Effects (PFE) and a frontier Spatial Autoregressive (SAR) model on a unique panel dataset for 375 local authorities in England and Wales between 2003 and 2018. Results from the analysis indicate that the presence of pubs progressively relates to a higher incidence of major crimes when transitioning from rural to urban areas, mainly due to weaker level of community cohesion and a lack of resources to support formal policing in more urbanized centres. These findings highlight the importance of place-based strategies in tackling rising incidences of crimes, indicating that recent pub closures may have contributed to severing community ties that act as a deterrent to crime in certain areas.

KEYWORDS:

• Pubs
• crime rates
• local authorities
• Great Britain

1. Introduction

In the United Kingdom (UK hereafter), public houses or pubs provide important places for social aggregation, offering physical settings for many types of communal activities (Maye, Ilbery, and Kneafsey 2005; Mount and Cabras 2016). Several recent studies confirm the positive impact of pubs, as third places, on local communities (Bowler and Everitt 1999; Maye, Ilbery, and Kneafsey 2005; Cabras and Bosworth 2014; Cabras and Mount 2017). Indeed, their positive impact on community cohesion is greater than other third places including community halls, village shops, post offices, and libraries (Cabras and Mount 2017). Pub closures therefore represents a potential threat to the cohesion and attractiveness of a community, but their number declined from about 67,000–46,350 in the period 1982–2021 (British Beer and Pub Association 2022)

Notwithstanding the importance of the issue and the significance of the long run decline, there is an absence of longitudinal large sample research evidence into the consequences of pub closures. One reason is that significant attention has concentrated on the causes of these closures, notably regulation, restructuring, and competition from supermarkets (Preece 2016). A further reason is the lack of a consistent long run data source suitable for panel analysis of the effects of closures on community cohesion and economic wellbeing. Moreover, because pubs offer generalized social benefits, placing an economic or social value on their presence or absence in a community is problematic. These benefits may differ according to access and location effects, for example according to the degree of urbanization.

A possible solution, proposed in this paper, is to explore crime rates across these locations to assess the effects of pubs and pub closures in any given locality. Access and proximity to amenities are important factors in households’ choice to purchase a house in the countryside (Costello 2007; Cabras et al. 2021), and in urban and suburban areas, particularly in congested metropolitan areas (Ding, Zheng, and Guo 2010). A survey from Tepilo (2015) indicate a substantial number of homebuyers specifying proximity to a pub as important in their decision, as pubs have a positive impact on house prices. In a similar vein, specifying the relationship between crime and pubs through time and across a range of rural/urban contexts would increase our understanding of the social consequences associated with the presence of these businesses at a local level.

To examine this relationship, we analyze an original longitudinal dataset comprising information on facilities and services available for 375 local authorities (LAs hereafter) in England and Wales between 2003 and 2018, grouped according to the degree of urbanization. We address the following research questions: what is the relationship between pubs and crime rates in England and Wales? And how is this relationship mediated between urban and rural LAs?

The paper comprises of five sections, including this brief introduction. Section two discusses and illustrates the theoretical background of the study, focusing on third places, community cohesion and social capital considering the progressive decline in pubs that has occurred in England and Wales. Section three describes methodology and data, explaining the econometric models elaborated and applied in the analysis. Section four examines results and discusses findings. Section five provides conclusions.

2. Background

2.1. Social cohesion, third places and the decline of pubs in the UK

There is a long-standing concern about how urbanization and industrialization affect the social fabric of communities (Bellair 2017). In local communities and neighbourhoods, informal measures of social control, mainly arising from social relationships forming among residents, might act as a deterrent for crime. Informal control could arise because of friendship, organizational, or other network ties including residents’ supervision of social activity within a neighbourhood, as well as the institutional socialization of children toward conventional values (Bellair 2017). This assumption is traditionally associated with concepts such as community or group cohesion, social integration, and trust (see Oldenburg 1989; and Putman 2000). The link between social cohesion, thus the degree of linkages within different social components within a community, and informal social control is explained by collective efficacy, a theory introduced by Sampson, Raudenbush, and Earls (1997) in a study of Chicago neighbourhoods. According to this theory, social cohesion within a given community activates collective action towards a common good, such as maintaining public order or preventing crimes (Sampson 2012). While it is generally accepted that urbanization tends to weaken community ties and create more individualistic societies, although collective efficacy rests on the premise that informal control in urban neighbourhoods could still exist if the neighbourhood is cohesive e.g.: its residents have shared values and trust (Sampson 2012).

In such a context, third places occupy an important role. Third places are defined as physical places outside households (first places) and working environments (second places) for people to congregate and join together. Third places shape frameworks and boundaries for individuals and groups (Oldenburg 1989; Watson and Watson 2012), and facilitate the accumulation of social capital within the communities they serve; with social capital defining the whole of relationships and ties among individuals which provide a degree of social interaction, cohesiveness and networking in a given community (Putman 2000). In the UK, empirical evidence suggest that pubs, as third places, play an important functional role in providing platforms for these concepts to develop and expand (Cabras and Mount 2017). Despite this evidence, however, the number of UK pubs has been progressively declining since the early 1990s (Cabras, Higgins, and Preece 2016). This decline has been caused by several factors, an important one represented by legislation introduced by Parliament in the late 1980s (the Beer Orders), which saw the separation of pubs from breweries that traditionally owned them, and the consequent rise and enlargement of corporate pub-chains or pubcos dedicated to retail (Cabras, Higgins, and Preece 2016). Other factors like the decrease of alcohol prices sold in supermarkets and off-licence retailers (Smith and Foxcroft 2009); the rise of theme pubs and European style-cafes (Lincoln 2006); the growth of home entertainment in terms of affordability of devices such as high-definition TVs and home-theatre sound systems (Cabras, Higgins, and Preece 2016); all contributed to making pub nights less attractive. More recently, the pandemic crisis and the high inflation mainly created by high energy price have further exacerbated the situation for pubs (Shakina and Cabras 2022).

In 2011, the vast majority of pubs fell under direct control of large national brewers and corporate pub chains to control approximately 55 percent of all pubs (Preece 2016). However, some of the largest pubcos incurred heavy losses during the 2008 financial crisis, later generating significant disinvestment, ownership changes and several pub closures (Andrews and Turner 2012). At the end of 2014, two fifths of British pubs were owned by pubcos, two fifths were free-houses and the rest owned by breweries (Cabras and Mount 2017). In July 2016, Parliament approved the Market Rent Only (MRO) option aimed at reducing the level of interference of pub landlords on their tenants and lessees, providing them with new rights and protections such as the ability to negotiate a fairer rent and to move ‘free of tie’ from their landlords, although landlords were still responsible for ensuring pub premises (Shakina and Cabras 2022). Since the introduction of the MRO option, the number of free houses in the country has increased. In 2020, about 50 per cent of the UK pubs were independently owned, while 21 per cent were owned by brewers and 29 per cent by pubcos, with large national pubcos still directly controlling about 13,900 pubs (Shakina and Cabras 2022).

Pubs face different challenges and issues depending on where they are located. In urban areas, pubs represent an important component of the so called ‘night- time economy’ (NTE), along with clubs, bars and other licenced premises that attract residents as well as tourists to urban centres (Hough and Hunter 2008). During the 1990s and early 2000s, the NTE was encouraged by successive Governments with licencing and planning policies, within a national strategy that sought the re-vitalization of city-centre locations alongside the development and expansion of retail and leisure attractions (Hough and Hunter 2008). Pubs located in rural areas did not enjoy similar support; many had to reconfigure themselves into different types of businesses (e.g. gastro-pubs, bed and breakfasts) in order to survive.

2.2. Determinants of crime at a local level

Several studies in literature investigated the relationship between crime and economic conditions (Elliott and Ellingworth 1998; Carmichael and Ward 2001; McIntyre 2017) and between crime and social contexts (Cantor and Land 1985; Agnew 1992). Evidence from these studies suggest a relationship between crime and unemployment, although there are no conclusive findings whether this relationship is positive (see Carmichael and Ward 2001; Reilly and Witt, 1992) or negative (see Cantor and Land 1985; Deadman and Pyle 1994).

Agnew (1992) proposes the General Strain Theory (GST) to explain how economic hardship could generate higher levels of crimes in societies. Three main components in the GST identify why individuals may be moved towards crime: when they fail to achieve their desired goals, when they lose positive values and, in contrast, they are exposed to negative valued stimuli. The combined presence of these three strains on individuals and communities can explain how events such as financial hardship and unemployment may result into crime (McIntyre 2017). Examining relationships between different types of crime and levels of personal debt default in London after the 2008 financial crisis, McIntyre (2017) finds that increases in debt default in neighbourhoods impose a significant social cost on communities through increases in theft crimes. Field (1990), analyzing crime rates in England and Wales, finds personal consumption levels as an important factor in determining crime rates, and points out that regional and local variations in economic conditions are frequently neglected in economic models measuring the association between personal consumption and crime.

Existing literature already shows that crime rates are unevenly dispersed and higher in urban areas (Weisburd, Groff, and Yang 2012), and rates vary across different times of the week or even day (Newton 2015). Moreover, the relevance, magnitude, and frequency of crime can vary according to type across locations and neighbourhoods (Hipp, Tita, and Boggess 2009). Ellen, Horn, and Reed (2019) demonstrates that reductions in crime in given areas can translate into increased property values and attractiveness for those areas, which may further determine a progressive change in neighbourhood population characteristics. The same happens when crime spurs: increased level of criminal activity have consequences for neighbourhood investment, although impact can widely differ across areas (McIlhatton et al. 2016). Examining crime levels across a diverse set of US cities, Acolin et al. (2016) find the spatial spillover effects of crime on private investment to be limited in most of the urban areas, and highly localized at the micro scale.

Personality of the individual, cultural transmission, type of social structure and related imbalances, norms and conventions are also important factors to consider when analysing crime (Goglio 2004). By excluding individual personality, however, it appears that the contextual factors indicated above exert significant influence on criminal behaviour and remain relatively stable in the social context in the short-medium term. Goglio (2004) identifies the important role played by institutions to understand the effects of economic development on criminal activity, stating that

(economic) development almost inevitably brings with it situations of dualism which weaken social cohesion if they are not corrected (…) If the institutions grow obsolete more rapidly than they are replaced, a dearth of culturally internalized general institutions may come about, with a consequent further erosion of social cohesion. (861)

In terms of institutions, the administrative structures and operational frameworks related to police forces play an important role in terms of crime prevention at a local level. Reforms occurred in the re-organization of local police authorities during the 1970s and 1980s in the UK brought England and Wales to experience a de facto ‘national’ police force (Reiner, 2013), characterized by some significant centralization in terms of administration, management, and funding. Local policing became more heavily micro-managed by the Home Office, with local police authorities driven substantially by objectives and directives set at a national level (e.g. the introduction of the ‘National Policing Plan’ and a Police Standards Unit to assess, monitor, and vet the performance of local forces; Edwards et al. 2017). The Home Office controlled the greater proportion of police funding, with most of the financial support for provincial forces coming from national government. The 1998 Crime and Disorder Act recalibrated the focus of police authorities towards a more localized dimension, promoting ‘Neighbourhood Policing’ by engaging with private, public and voluntary sector partners in view of understanding local communities and building trust, and introducing initiatives such as community safety partnerships and programmes in local authority areas (Jones and Lister 2019). Since the early 2000s, however, such partnerships and programmes of ‘localized’ policing have been frequently affected by budgetary and policy-making constraints: for instance, key decisions about resource allocation and deployment among local police forces were still made at higher levels, limiting police authorities in terms of generating innovative response to ‘local’ problems. Moreover, the development of multiple regional collaborations across local police forces to share operational resources over the years appears to have increased the level of centralization by members of more ‘localized’ policing networks (Reiner, 2013).

2.3. The relationship between crime and pubs

The relationship between pubs and crime is complex and multifaceted. As third places, pubs can facilitate a sort of informal control that could arise as a result of friendship, organizational, or other network ties which include residents’ supervision of social activity at local level, similar to the institutional socialization of children toward conventional values (Bellair 2017). In relation to pubs, existing studies demonstrate the presence of an informal reputational system prevented the appearance of antisocial behaviours, with publicans often working together with residents in addressing issues related to excessive drinking (see Cabras and Mount 2017). Research also identifies difference in perception between men and women in the value of pubs for their own communities, particularly regarding antisocial behaviours, with women tending to indicate pubs as places where antisocial behaviour could happen more frequently compared to other places (Leyshon 2008). More generally, analyzing the association between third places, drunkenness and antisocial behaviour across different drinking geographies requires attention to different components living within the same communities (Jayne and Valentine 2016).

Particularly in rural areas, pub closures generated a general change in local customers’ behaviour, with some residents pushed to drink at home and entire sections of the community, for instance farmers living in isolated hamlets and farms, being penalized most (Cabras and Mount 2017). The risk of isolation associated with drinking at home appeared also to be perceived differently: older residents and (to a lesser extent) women might be more concerned on the effects home drinking was having on social relationships among individuals outside family ties, while views expressed by younger participants and men in general tended to focus on issues occurring within families. These different perceptions expressed by residents resemble findings from other studies (Laoire 2001; Jayne and Valentine 2016).

Previous studies (Cabras and Mount 2017; Cabras et al. 2021) value the role pubs play within communities in relation to fostering and supporting socialization processes, although again perceptions may vary across different components in the same community. For instance, self-employed and retired residents may tend to praise the direct support of pubs and publicans to the local economy while, for younger residents, publicans appeared essential for the establishment and sponsoring of clubs or sport teams, whose presence in the areas would be extremely reduced, if existent at all (Everitt and Bowler 1996; Cabras and Bosworth 2014).

3. Methodology

3.1. Data

To address our research questions, we extracted and combined a range of information from various datasets provided by the Office for National Statistics (ONS 2019), including those provided by Annual Business Enquiry (ONS 2019). Data collected encompasses a time-period between 2003 and 2018. The unit of analysis in this study is the local authority (LA); data in each LA comprises of number of pubs operating in any given year, house prices, unemployment rates, resident population, and crimes recorded in nine categories – criminal damage and arson; all vehicle offences; public order offences; weapon offences; sexual offences, stalking, and harassment; violence with injury; violence without injury; and thefts. Detailed descriptions of variables are provided in the appendix (see Table A1).

LAs considered in our study were further classified according to their levels of urbanisation/rurality using the classification by Bibby and Shepherd (2004).¹ Six main categories are identified: ‘Major Urban’ (districts with a resident population of 100,000 people or more, with 50% of the population concentrated in urban areas comprising more than 750,000 residents); ‘Large Urban’ (districts with a population of 50,000 people or more, or with 50% of the population concentrated in urban areas showing between 250,000 and 750,000 residents); ‘Other Urban’ (districts with a fewer than 37,000 people, or with at least 26% of their population in larger market towns and rural settlements); ‘Significant Rural’ (districts with more than 37,000 people, or more than 26% of their population in larger market towns and rural settlements), ‘Rural – 50’ (districts with between 50% and 80% of their population in rural settlements and larger market towns); and ‘Rural – 80’ (districts with at least 80% of their population in rural settlements and larger market towns). To conduct our analysis, we condense the six consolidated categories into three geographical settings: Mostly Urban (comprising ‘Major and Large Urban’); Semi Urban (comprising ‘Other Urban’ and ‘Significantly Rural’); and Mostly Rural (comprising ‘Rural 50’ and ‘Rural 80’).

We use Bibby and Shepard’s urban-rural classification as it addresses areas at the same administrative level across the two countries considered, and for which there is an appropriate amount of data available for the period analysed. Equally, we decide to focus on LAs as these administrative units provide a much clearer distinction in terms of scalar levels of urbanity/rurality compared to instead of other administrative or geographical units (see Cabras and Mount 2017). Based on these considerations, we apply the urban-rural classification to examine and model interaction effects in view of analysing how the relationship between pubs and crime rates varies across LAs with different levels of urbanization/rurality.

3.2. Descriptive statistics

To examine the relationship between pubs and crimes, we start by comparing their numbers across years and different LAs classified as mostly rural, semi-urban and mostly urban. Figure 1 shows a marked decline in number of pubs across all three classifications since 2008, with the highest number of pubs situated in Mostly Urban areas, followed by Mostly Rural, and then Semi Urban. Along with the decline in the number of pubs, we observe a steady rise in the number of crimes which is in line with previous findings of d’Orban (2021).² Figure 2 demonstrates that number of sexual offences, violent crimes with and without injury, and cases of public disorder increased over 2003–2018 time period. The same trend is observed for all other crime type (see Figure A1 in the appendix),³ particularly in urban areas, corroborating evidence that crimes are unevenly dispersed towards regions of high urbanity (Weisburd, 2015).

Figure 1. Number of pubs in Great Britain 2003–2018.

Figure 2. Crime trends 2003–2018.

Figure 3 shows that at least half of all crimes happen in Mostly Urban areas; up to a third of all crimes take place in Semi Urban areas; whereas less than a quarter of crimes are committed in Mostly Rural areas. The three most prevalent crimes in Mostly Urban areas are vehicle, theft, and public disorder offences, whereas criminal damage, stalking and harassment, as well as cases of violence are the leading crimes in Mostly Rural and Semi Urban locations. These findings corroborate previous studies indicating people living in rural areas at a lower risk of facing a crime than people living in urban and inner-city areas (Pateman 2011), and that non-rural households are three times more likely to be targeted for theft of vehicles (Marshall and Johnson 2005). Also, they confirm a wider spread of public disorder in urban locations (Hallsworth and Brotherton 2011), as well as a high frequency of crimes related to harassment and criminal damage in rural zones (Garland and Chakraborti 2006). While crimes with weapons, sexual offences, and violent crimes with injury appear to be significantly related with the number of pubs only in Semi Urban and Mostly Urban locations. Overall, a high concentration of pubs in urban areas is complemented by a large number of crimes, although there is a marked variation in types of crimes depending of the level of urbanity/rurality.

Figure 3. Crime incidence by location.

Table 1 presents the pairwise Spearman correlations between number of pubs and incidence/type of crime across the three locational settings. Theft, criminal damage as well as vehicle offences are positively correlated with the presence of pubs across all area types, corroborating previous studies indicating vehicle stealing and damaging as common on the streets with places for social gatherings like pubs (Beck and Willis 2006; Kinney et al. 2008); as well as bags theft widespread in pubs and bars (Sidebottom and Bowers 2010). The table shows a positive association with the number of pubs and crimes in Mostly Rural areas, but it is not significant. On the contrary, in urban areas and semi-urban areas, lower levels of sense of belonging and social responsibility, along with minimal ties within local communities, could contribute to higher incidence of crimes.⁴ Antisocial behaviour and violent crimes appear to be significantly correlated with the number of pubs in urban locations only, not with pubs in rural areas, sustaining the hypothesis that these places might work as informal centres of policing and monitoring in the countryside (Mount and Cabras 2016; Cabras and Mount 2017).

Table 1. Spearman correlations between number of pubs and incidence of crime.

	Mostly Rural	Semi Urban	Mostly Urban
Violence w/ injury	0.0751	0.2723***	0.2743***
Violence w/o injury	−0.0892	0.112***	0.0182
Stalking and Harassment	−0.0993*	0.0329	−0.1112***
Sexual Offence	−0.08	0.1348***	0.1298***
Public Disorder	0.0659	0.1929***	0.1983***
Weapons Offence	0.0736	0.2075***	0.2614***
Theft	0.2886***	0.2219***	0.1968***
Criminal Damage and Arson	0.257***	0.3654***	0.4956***
Vehicle Offences	0.2822***	0.3691***	0.3113***
Observations	304	1136	976

^∗ p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001.

3.3. Econometric modelling

We started our analysis by applying a LA fixed effects specification to control for time invariant unobservable factors at the LA level (results in appendix). However, spatial fixed effects are not sufficient in view of identifying and assessing the presence of spatial interactions such as spillovers associated with the value of a dependent variable  – incidence of crime in our case  – in one location, to values observed at other (nearby) locations. This task can be accomplished by assigning spatial weights within a Spatial Autoregressive (SAR) empirical specification.

Spatial models have been applied in a variety of disciplines, such as criminology, demography, economics, epidemiology, political science, and public health. Following Lee and Yu (2010), we estimate the following fixed effects spatial panel data model: $\begin{array}{r}{\mathrm{y}}_{\mathrm{it}}={\mathrm{X}}_{\mathrm{it}}\text{β}+\lambda {\text{Wy}}_{\mathrm{it}}+\gamma {\text{WX}}_{\mathrm{it}}+{\mathrm{c}}_{\mathrm{i}}+{\mathrm{u}}_{\mathrm{it}}\\ {\mathrm{u}}_{\mathrm{it}}=\rho {\mathrm{Wu}}_{\mathrm{it}}+{\text{v}}_{\mathrm{it}}\mathrm{t}=1,2,...,\text{T}\end{array}$

The model allows for incorporating spatial lags of the dependent variable, the explanatory variables, and the errors. We then implement the quasi-maximum likelihood (QML) estimator to fit the model (see Lee and Yu, 2010). A transformation is used to eliminate the fixed effects, generating the following specification: $\begin{array}{rl}& {\tilde{\mathrm{y}}}_{\mathrm{it}}={\tilde{\mathrm{X}}}_{\mathrm{it}}\text{β}+\lambda \mathrm{W}{\tilde{\mathrm{y}}}_{\mathrm{it}}+\gamma \mathrm{W}{\tilde{\mathrm{X}}}_{\mathrm{it}}+{\tilde{\mathrm{u}}}_{\mathrm{it}}\\ & {\tilde{\mathrm{u}}}_{\mathrm{it}}=\rho \mathrm{W}{\tilde{\mathrm{u}}}_{\mathrm{it}}+\tilde{\mathrm{v}}\end{array}$

where y_it =  (y_it,y_2t, …, y_nt) is an n × 1 vector of observations for the dependent variable for time period t with n number of panels; X_it is a matrix of time-varying regressors; c_i is a vector of panel-level effects; u_it is the spatially lagged error; v_it is a vector of disturbances and is independent and identically distributed (i.i.d.) across panels and time; and W is the spatial weighting matrix for UK local authorities.

In our model, the dependent variable is the mean incidence of a specific crime type in LA i at time t. The different types of crimes investigated are criminal damage and arson, vehicle offences, public order offences, weapons offences, theft, sexual offences, stalking and harassment, violence with injury, and violence without injury. The independent regressors are: Pubs_it, the main variable of interest constructed as the number of pubs in LA i at time t; interacted with SemiUrban_it, a binary variable that specifies whether the LA i is in a Semi Urban⁵ area, and MostlyUrban_it, a binary variable that specifies whether the LA i is in a Mostly Urban area; with Mostly Rural assigned as the base level.

In our analysis we also apply a set of controls at LA level comprising unemployment rate, average housing prices and population. Housing prices are used to capture the level of income that inhabitants of a certain area might have and thus provide an estimation of residents’ prosperity/poverty, which might play an important role as the literature shows the significant causal effect of poverty on crime (Valdez 2007; Iyer and Topalova 2014). Unemployment is also included in our regression model since the literature evidence that unemployment has a positive and significant effect on crime occurrence (Edmark 2005; Fougère, Kramarz, and Pouget 2009). The weighting matrix W is effectively a constraint placed on the individual spillovers formulated as part of the model specification. We estimate with two types of spatial weighting matrix: one where weights represent spillovers between adjacent local authorities (called ‘contiguity’ in the literature); and another where spillover effects are proportional to the inverse of distance between local authorities. Results are nearly identical in both cases.

As coefficients in the SAR model cannot be interpreted directly, we estimate marginal effects, thus the effect of number of pubs on incidence of a specific crime type in each of the location categories. The marginal effects are presented in Table 2. The full set of estimation results of the SAR model are presented in the appendix.

Table 2. Marginal Effects of number of pubs on crime.

(a)
	Criminal Damage and Arson	Weapons Offence	Public Disorder	Theft	Vehicle Offences
Mostly Rural	72.585	0.648	14.010	24.589***	57.133
	(40.897)	(0.857)	(7.899)	(5.912)	(29.487)
Semi-Urban	101.372***	0.437	8.135	35.165***	117.342***
	(30.053)	(0.646)	(5.848)	(4.680)	(22.176)
Mostly Urban	80.315**	−0.167	5.623	18.436***	61.206**
	(28.024)	(0.619)	(5.600)	(3.704)	(19.478)
N	2416	2416	2416	2416	2416
Chi-Squared	552.824	391.934	715.682	1244.231	642.965
Pseudo R-Sq	0.086	0.479	0.240	0.204	0.123
(b)
	Sexual Offence	Stalking and Harassment	Violence without injury	Violence with injury
Mostly Rural	−11.332***	−15.650***	−34.745***	−4.959
	(1.693)	(4.747)	(7.117)	(10.431)
Semi-Urban	−11.891***	−20.694***	−37.432***	−5.081
	(1.247)	(3.512)	(5.251)	(7.932)
Mostly Urban	−10.479***	−21.675***	−31.318***	−5.879
	(1.196)	(3.379)	(5.044)	(7.654)
N	2416	2416	2416	2416
Chi-Squared	2623.193	1446.613	2129.134	823.126
Pseudo R-Sq	0.397	0.261	0.313	0.469

Standard errors in parentheses *p < 0.05, **p < 0.01, ***p < 0.001.

4. Results

The marginal effects for each of the rural-urban category must be interpreted in exclusive terms, e.g.: the marginal effect against mostly rural areas shows the association between change in number of pubs and the mean incidence of that crime type in Mostly Rural areas only; the same applies for Semi Urban and Mostly Urban areas. In Table 2, the marginal effect of an increase in number of pubs by one is an increase in theft offences by 24.589 in Mostly Rural areas, 35.165 in Semi Urban areas, and 18.436 in Mostly Urban areas, with all the estimates statistically significant. Similarly, the increase in number of pubs by one is associated with a decrease in sexual offences by 11.332 in Mostly Rural areas, 11.891 in Semi Urban areas, and 10.479 in Mostly Urban areas, with all the estimates statistically significant. The estimated marginal effects presented in Table 2 confirm that the association between presence of pubs and incidence of crime highly depends on the area type as well as the crime type.

Overall, we find that the marginal effect of pubs on crime is either positive and statistically insignificant, or negative and statistically significant for Mostly Rural areas (with the exception of theft offences). This lends support to the anecdote that, in rural areas, social cohesion and the resulting informal reputational and control system makes pubs a place of crime prosecution and policing. The picture is different for Semi Urban areas, where we find a positive and statistically significant effect for three crime categories – criminal damage and arson, theft, and vehicle offences. The contrast with estimates for Mostly Rural areas indicates that the dynamic between pubs and crimes might differ in relation to this locational category. The coefficients for Mostly Urban areas are very similar to those associated with Semi Urban areas, albeit smaller. For all locational categories, the association between pubs and crime for sexual offences, stalking and harassment, and violent crimes without injury is negative and statistically significant.

Criminal damage and arson, stalking and harassment, and cases of violence without injury remain the most common in Mostly Rural areas, and they either have a negative association with the presence of pubs or no significant associations at all. In contrast, Mostly Urban areas show a high number of vehicle offences, public disorder, and theft, and the presence of pubs in those areas has a positive and significant (except public disorder) association with these crimes. The most frequently occurring crimes in Semi Urban areas are the ones that show a negative association with number of pubs in these locations: cases of violence with and without injury are not only among the most common crimes in Semi Urban areas, but seems to be negatively associated with the number of pubs within spatial proximity.

Overall, we find no evidence that pubs contribute to increased incidence of crimes in rural locations and instead are mostly associated lower incidences of crimes; as the level of urbanization increases, pubs seem to have a mixed impact depending on the type of crime.

5. Discussion

A possible underlying mechanism that helps explain our findings is the weakening of informal social control, which acts as a deterrent for crimes at a local level (Bellair 2017), in the transition from rural to urban locations. Results seem to confirm that, in rural areas, social cohesion and the resulting informal social control arises as a product of a tight knit community with shared values, similar family structures, and similar ethno-religious backgrounds (Watson and Watson 2012). Pubs, as third places, act as a conduit for fostering community cohesion and hence are much less likely to be epicentre of anti-social behaviour and crimes in rural areas, and a high degree of social cohesion lends itself to preventing crimes through informal social control (Cabras and Mount 2017).

In rural areas, active communication and cooperation between publicans and the police, mostly to prevent violent and antisocial behaviour on their premises, seems to work as a powerful deterrent for criminal activities (Cabras and Mount 2017), confirming the importance of community and informal policing when dealing with crimes (Renauer 2007; Anderson and Britain 1997). Moreover, the smaller population of rural communities and a higher degree of acquaintanceship increases the level of awareness of residents towards crime, helping to lower their crime rates (Cabras and Mount 2017). This can possibly explain why much lower levels of crimes are observed in rural areas (even after controlling for population) than in urban areas. Hence, caution is needed when drawing links between the presence of pubs and incidence of crimes in rural areas, as shown by our results for Mostly Rural areas.

While pubs do not significantly affect the incidence of most major crimes in Mostly Rural areas, results paint a distinctly different picture for Semi Urban and Mostly Urban areas. Our results indicate that positive and statistically significant associations between presence of pubs and offences of criminal damage and arson, theft and vehicles are largest in Semi Urban areas. These marginal effects are similarly positive and statistically significant but smaller for Mostly Urban areas. In our analysis, Semi Urban areas experience higher levels of urbanization than rural areas, although they would still be economically behind highly urban areas. Semi Urban areas are also more likely to consist of a heterogeneous population from vastly different ethno-religious and socio-economic backgrounds; many of their residents could potentially have migrated out of rural areas in search of employment in urban centres, or for higher education (see Acolin et al. 2016). As a result, these neighbourhoods may lack the strong social ties, kin networks, friendships, and trust prevalent in smaller rural communities, and hence are likely to have lower levels of informal social control. Moreover, they are likely to experience lower levels of policing than highly urban communities given the economic status of residents, lower council tax bands, lower income, and employment. Overall, this can reduce or even erase formal and informal controls and, in such a context, pubs could potentially provide hotspots for criminal activity, especially if the types of crime are linked to alcohol consumption.

The link between pubs and crime in Semi Urban areas is mediated by the lack of community cohesion on one hand and formal policing on another, meaning that simply removing or restricting access to alcohol without complementary policy measures to deter crime might not lead to safer neighbourhoods. Instead, more effort should be devoted to improving policing within Semi Urban neighbourhoods with initiatives aimed at strengthening community ties and cohesion, e.g.: promoting events such as markets, fairs and exhibitions, or cultural initiatives aimed at improving engagement and participation to communal activities among residents.

Our findings indicate that for crimes of a more personal nature, such as sexual offences, stalking and harassment, and violence with or without injury; the presence of pubs does not exacerbate the incidence of these crimes but instead shows a negative association. These findings are important in view of tackling and better addressing criminal behaviours by effectively enhancing the role of third places such as pubs within local communities and neighbourhoods in England and Wales, dealing with specific types of crimes. For instance, in March 2023, the UK Government launched the ‘Anti-Social Behavior Action Plan’ with the objective of bringing different partners and stakeholders together in clamping down anti-social behaviour (HM Government, 2023). The plan sets out a new framework for ‘the Government, police forces, Police and Crime Commissioners, LAs and other partners – such as housing associations and youth offending teams – to work together to address the many drivers of anti-social behaviour and repair the damage to communities’ (2023, 7). Notably, third places such as pubs or community halls, which help facilitating the establishment of individual relationships and socialization processes, are not mentioned.

The same plan acknowledges a range of factors influencing the willingness of people to report anti-social behaviour, the importance of acquiring good data at a local level to take effective action and improve people’s lives. Empirical evidence already identified the positive role played by pubs and publicans in monitoring and preventing crime such as anti-social behaviour within rural areas (e.g.: Cabras and Reggiani 2010; Mount and Cabras 2016). This is mainly due to the instauration of a ‘reputational system’ which works as a deterrent for residents to engage in inappropriate activities, in which pubs function as main centres for local knowledge and raising collective awareness among residents about multiple issues happening within their spatial proximity (Cabras and Mount 2015). Likewise, investing to expand the relevance and centrality of third places such as pubs within urban and semi-urban areas in England and Wales could bring multiple benefits in terms of acquiring better information as well as monitoring and preventing crime.

6. Conclusions

The aim of the paper was to analyze the relationship between pubs and crime rates in England and Wales, exploring how this relationship mediated between urban and rural LAs. The econometric models developed on data collected from 375 LAs in England and Wales between 2003 and 2018 identified a range of associations between presence of pubs and different types of crimes recorded in the period considered, some of these associations resulting statistically significant. In rural areas, marginal effects related to the incidence of crimes on pubs’ density identified a significant relationship between pubs and crimes classified as public disorder, while in urban areas the analysis identified a significant relationship between pubs and crimes classified as violence without injuries and, to a lesser extent, damage and arson. The incidence of crime in relation to pubs was higher in sub-urban areas, with strong relationships identified for crime classified as theft and sexual offences.

The analysis yields three important results. Firstly, the analysis indicates that rural LAs experience a much lower incidence of crime issues related to the number of pubs compared to more urbanized LAs. While this finding corroborates those from previous studies (Renauer 2007; Anderson and Britain 1997; Mount and Cabras 2016) about the positive role played by pubs within rural communities, it enhances and expands their relevance with fresh quantitative evidence. Secondly, our econometric models identify a positive relationship between the number of pubs and specific types of crimes depending on level of urbanity/rurality, supplying evidence in view of designing and developing policies and initiatives aimed at increasing the effectiveness of policing at local level and, at the same time, increasing levels of community cohesion and trust within local communities (Cabras and Reggiani 2010; Mount and Cabras 2016). Thirdly, our analysis hints that the decline in the number of pubs in UK, observed in our data regardless of geographical areas, could potentially have significant repercussions on monitoring and understanding crime levels at a local level. As recent UK government initiatives tackling crimes neglect the role pubs and publicans play within local communities, our findings provide insights into how these places can be effectively used to achieve targets in terms of crime control.

While our study addresses an original topic, bringing new knowledge about the relationship between crime and third places at a local level, we acknowledge some limitations. Firstly, the econometric models are based on a fifteen years’ time span which limit our analysis just before the Covid-19 outbreak in 2020, which is likely to have had an impact on crime rates and pubs’ businesses, although we could not find any consistent information related to the data captured in our analysis from 2019 onwards. Secondly, and in part related to previous comment, our study focuses on data from England and Wales without expanding its analysis to other British countries or regions. Again, the lack of consistent information prevented us from extending our analysis from a geographical perspective: for instance, the way crimes are classified and recorded in Scotland differ significantly from the rest of the UK; the same consideration applied for criminal records in Northern Ireland. Thirdly, we are aware that the three spatial macro-dimensions analysed in our study provide a general overview of what happens in urban, suburban and rural areas: while these three categories enabled us to capture main trends affecting both pubs and crime rates surveyed in our study, it also neglects some notable exceptions (e.g.: pubs located in significantly rural LAs based in large conurbations, with related impact on crime analysis). Lastly, although our econometric models have been developed in a rigorous manner, they provide a broad but limited overview of the themes and issues addressed in our investigation. Time and financial constraints related to this study project prevented us from acquiring and adding more qualitative information, for instance, via means of interviews or focus groups, which would have expanded the quality of our findings even further.

In conclusion, our findings provide supporting evidence towards the importance of place-based strategies in tackling social issues such as crime levels in the presence of pubs. As the association between crime and pubs is multi-faceted, the formation of formal and informal policing (or lack thereof) gravitating around pubs plays a crucial role in whether the presence of pubs acts as a deterrent or an enabler of criminal behaviour. Given the progressive decline in the number of pubs in England and Wales, and generally in the UK, and the paucity of studies targeting the effects of this decline on local communities, this paper provides a fresh and timely contribution to the literature. It also provides a venue for new research aimed at better understanding the potential of pubs in terms of generating social capital, strengthening community cohesion, and preventing crime.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 This classification is currently applied to LAs in England.

2 A big rise in the number of crimes from 2006 to 2007 is associated with changes in approaches to crimes’ recording and classification (Berman 2008). Overall, the methodology of data collection and recording has been changing over the last 20 years. Given that, we do not focus on crimes in specific years, but rather look and compare their overall dynamics with the dynamic in the number of pubs from 2003 to 2018.

3 We do not place all the graphs in the main body of the paper for brevity.

4 We observe a spike in incidence of crime across all types in 2006 due to the change in classification and recording (Berman 2008). In order to avoid any spurious associations, we also estimate Spearman correlations for the dataset truncated to years 2007-2018. There are no major changes to any of the correlation estimates (see Table A2 in the Appendix A).

5 Geographic location is categorized into Mostly Rural, Semi Urban, and Mostly Urban.