Crime in Border Regions: the Scandinavian Case of Oresund, 1998-2001

Crime in border regions: The Scandinavian case of Öresund, 1998-20011

Vania Ceccato Urban Studies Division Royal Institute of Technology Drottning Kristinas väg 30 10044 Stockholm Sweden

Robert Haining Department of Geography University of Cambridge Downing Place Cambridge CB2 3EN England

1 This paper was presented at 33rd Annual Conference, Regional Science Association International British and Irish Section, 20th to 22nd August 2003. This is the original version of the article that will be published in December 2004 by Annals of the Association of American Geographers. Here will you find the final article: http://www.ingenta.com/journals/browse/bpl/anna 2

Abstract This paper compares offence patterns at two points in time in Öresund, a Scandinavian border region that links Sweden and Denmark. The aim of the analysis is to contribute to a better understanding of the relationships between crime and demographic, socio-economic and land use covariates in a border area that has been targeted with long- term investments in transport – specifically, through the construction of a bridge – the Öresund bridge. The changes effected by the construction of the bridge might be expected to impact both on the levels and the geographies of different offences. The basic hypothesis here is that the bridge and regional structural changes as a result of its construction, will generate new patterns of offences by creating new sites for offending and new, more vulnerable, transient groups unfamiliar with the levels and types of risks in particular areas. The paper focuses initially on identifying changes in selected offences in the region and, secondly, on explaining changes in the regional geography of crime using aggregated offence data for the largest Swedish municipality in the region before and after the bridge was built. Spatial statistical techniques and GIS underpin the methodology employed. The paper concludes with a discussion of results and suggestions for future research. This paper is a contribution to a relatively underdeveloped area in the geography of crime and one that is likely to be of growing interest in Europe as a result of EU enlargement and increasing intra-European cross-border movement facilitated by improved communication systems.

1. Introduction

Regions in the European Union, during the last decade, have been targets of massive transport investments to promote social cohesion, diminish the economic disparities between countries and improve connections between European capitals (EC 1997, EC 1998). This increase in accessibility makes cities, regions and countries more vulnerable to crime through new patterns of mobility. Especially in border regions, places become more vulnerable to potential threats, generating as Ruggiero (2000) suggests, ‘undefinable fears’.

The literature on the impact of new transport links on crime patterns is comparatively sparse; few studies deal with crime patterns at or across international borders. A notable exception is the study of the Euro tunnel that links England to France. According to the Rikskriminalpolisen (1999), there is evidence that the transport link has changed the dynamics of crime, especially in areas near the borders. This paper contributes to a better understanding of the interactions between offences and demographic, socio-economic and land use covariates in Copenhagen and Malmo in the region of Öresund (Figure 1). This region has received large-scale investment in transport – specifically, through the construction of a bridge. We first present an overview of the geographical shifts in 3 offences between 1998 and 2001 and then seek to explain changes in the geography of crime using spatial statistical techniques and GIS in the Swedish border municipality of Malmö.

The Scandinavian border region of Öresund has been chosen for several reasons. First, in the Nordic context, Öresund is an important region since it has been moving up the European hierarchy in terms of growth, competing with other Northern European metropolitan areas like Berlin, Hamburg and Stockholm (Mattthiessen 2000, 2002). Substantial steps towards the region’s economic integration were taken through large-scale long-term infrastructural and business investments. After decades of discussion, the Swedish and Danish parliaments decided in 1993 to construct a fixed link between the two countries. The Öresund bridge, which opened in July 2000, is the first fixed link between Sweden and Denmark and is the most tangible indicator of the region’s spatial integration. The bridge combines a four-lane motorway with a dual track railway and both are connected to their respective national networks with customs offices at each end. The Öresund bridge consists of a bridge from Malmö (Sweden) to an artificial island in the middle of Öresund, and from there it is connected to Denmark by a tunnel to a peninsula located close to the Copenhagen metropolitan area. Before the bridge, Öresund was linked by boat traffic and the effect of the new bridge has been to produce a significant increase in cross-border flows with the creation of new sites where travellers make the crossing.

Figure 1 – The Öresund region. Source: Adapted from SCB, Orestad: statistik om Öresund, 2003.

Second, Baltic and Nordic regions play an important role as receptors and transit territory in international organised crime (Galeotti 1995), and therefore it is reasonable to expect that a new transport link could potentially affect criminal activities in the region. As Ulrich (1994) states, Baltic and Nordic regions are of value to organised crime, particularly in the east mainly because of their geographical proximity and the fact that they offer a new market. They also have modern communication networks; liberal consumer markets; rapid access to global financial markets via Nordic banks; easy accessibility for transhipment of goods or people to Western Europe via its customs-free and open borders; and the potential for establishing companies as potential fronts for illegal activity. Appendix 1 provides more detailed information on some of these points. 4

Finally, data availability for the region was a decisive factor in choosing Öresund. Both in Sweden and Denmark crime data is systematically recorded by the local Police authorities at a very detailed level of time and space. However the fact that offence data are as in most countries regarded as ‘very sensitive’, the process of acquiring and standardising the datasets was time consuming, difficult and costly (See Appendix 2)

The structure of this paper is as follows. The relationship between crime and communication systems in border regions is discussed in section two. This is followed in section three by discussion of a conceptual model for examining the impact of improving cross-border communication on crime patterns. The analysis of changes between 1998 and 2001 over the whole region are reported in section four. Section five discusses changes in the geography of crime by using cluster detection techniques in the Swedish city of Malmö. Section six presents the results of modelling the changes in offence patterns in Malmö followed by a summary of the results and directions for future work in section seven.

2. Literature overview

2.1 The relationship between crime and communication systems

Investments in transport infrastructure are realised in space through new elements of the landscape such as highways, roads, stations, airports, ports, and bridges. Long before the establishment of modern public transportation systems Colquhoun (1800), showed how transport systems shaped crime patterns and he described how thieves preyed on shipping in the Port of London. Ports and terminals were the hot spots of a massive number of crimes triggered by the movement and assemblage of valuable goods. For other examples of the impact of transport systems on criminal activities in 19th Century England, see (Tobias 1967). Today, transport links (e.g., highways, roads, bridges) and nodes (e.g., stations, airports), together with the related consumer services they provide, create areas of exceptionally high, short-term population turnover. Large numbers of travellers, moving through areas that they are not familiar with are particularly vulnerable to the activities of opportunistic motivated offenders.

A new transport infrastructure may have a direct effect on crime patterns by creating new sites for offending or altering patterns of routine activity by motivated offenders. According to Brantingham and Brantingham (1981) offences occur where criminal opportunities intersect with areas that are cognitively known to a motivated offender. Tremblay and Tremblay (1998) show, for instance, how public transportation create spaces that bring together different social groups, attract motivated offenders and create opportunities for crime. Loukaitou-Sideris (1999) reports several studies on transit crime, mostly in rail systems and subway stations (Block and Davies 1996; LaVigne 1997; Levine and Wachs 1986; Poister 1996).

Contracting space-time through the establishment of a new transportation link makes people more mobile. However, an increase in people’s mobility per se does not trigger 5 crime. The easy access to places may generate new patterns of offences by creating new sites for offending in areas that despite being crowded places, ‘lack capable guardians’. The concept of ‘capable guardian’ was initially used in a neighbourhood context and refers to a person who discourages crime from taking place, sometimes just by their presence (Cohen and Felson 1979). A more modern view of the concept considers a capable guardian as either a person or a thing that discourages crime from taking place. ‘Guardians’ may be formal (e.g., security guards or CCTV cameras) or informal (e.g., neighbours, (Office 2003)). Public spaces, like train stations and airports, suffer a double disadvantage. They are mostly equipped with impersonal surveillance (a CCTV camera is not useful if it is set up or sited wrongly) and other forms of surveillance are often ineffective (staff might be present in a shop, but may not have sufficient training or awareness to be an effective deterrent; or may not be willing to put themselves at risk). Travellers who could in theory be considered as informal guardians may in practice also be ineffective. Most travellers have no sense of ownership in places like airports and often do not want to get involved; to intervene during the act or later as witnesses. Therefore, transport nodes of any kind are typical examples of poorly guarded places. The same reasoning could be used to explain offence patterns that occur during an actual journey, on a bus or train. Moreover, an increase in the number of travellers intensifies pressure on public transport systems leading to crowding and congestion. This combined with any physical deterioration of the bus or railway station (e.g., poorly lit areas) may increase the risk of certain types of offences.

There is also an indirect impact of transport infrastructure on crime patterns. One example is through the creation of new vulnerable groups, such as tourists in environments that may offer little protection. The implementation of a new transport infrastructure often leads to other types of investments that generate changes in land use, such as the creation of a new shopping area. As Wikström (1991) states, land use determines both the activities found in an area and the composition of the population at any given time, therefore, spatial variation in land use affects spatial variation in the number of interactions that are criminologically relevant in the sense that they could lead to offences. Shopping areas attract temporary populations and thus they are at high risk from so-called ‘tourism-criminality’ (Rikskriminalpolisen 1999) such as, pick pocketing, shoplifting, thefts of different kinds, and in certain cases, violence. Travellers, unfamiliar with the levels and types of risks in particular areas, may be the perfect targets. In other cases, these new groups are the ones that see the chance to offend and get involved in opportunistic acts of violence and disorder (Dunning 2000), (Stott, Hutchison, and Drury 2001). Central areas are examples of where opportunistic acts of violence and disorder become concentrated, especially close to entertainment places and restaurants. More than half of all violent crime takes place in public places (close to restaurants and pubs), in or close to inner city areas (Wikström 1991). The establishment of out of town retailers, outlets, and convention centres, as part of new investments in the area stimulated by the bridge and the induced economic activity, could also generate new offence patterns.

2.2 Crime at a political border There are certain factors that make political borders and the areas close to them unique places for criminal activities (Table 1). Many of these factors are of a structural and 6 organisational nature relating to the states and not properties of the border itself. Nevertheless it is in the border where these factors become especially important as triggers of crime.

The border is uniquely exposed to potential external ‘adversities’2. In the law enforcement context, borders are associated with controls. Loss of these controls means that the border as an instrument of law vanishes. Despite the fact that frontier controls have become more relaxed within European countries, frontiers remain symbols of state sovereignty. In criminal matters, a border defines “the limits of (the) executive police powers (and) states are reluctant to loosen their grip on these powers since territorial sovereignty represents an important political and psychological threshold for any European country” (Anderson et al.. 1995:124). As Jamieson (1998:225) suggests, in an age of insecurity born of globalisation, “borders take on a new significance in the popular mind as a symbol of community well-being and security”, a border becomes “a kind of cordon sanitaire in a troubled, globalised world”.

However, borders also mean opportunities for crime. Hajdnijak (2002) points out how weakened state apparatus and political and social instability create opportunities for crime. In the former Yugoslavia ‘state borders have become the main mechanism for redistribution of national wealth’ (p.69). This is because a significant part of the population has actively been involved in criminal schemes, largely tolerated by the authorities. The borders represent a form of ‘safety valve’ for pressures created by the ever-increasing army of the unemployed. In these countries, ‘the border’ provides, according to Anderson et al., (1995), an alternative regime of social security that together with the weakened state apparatus provide new opportunities for illicit trade3.

When a new type of transport link between two neighbouring countries redefines the physical line of contact between them, then changes will occur in the activities at the border. Improvements in a region’s infrastructure, such as, from a ferryboat service to a bridge, facilitate interactions of people and goods. This in turn affects a region’s dynamics in terms of its economy, environment, people’s life styles, and not least, conditions for crime (see, for example, Andersson and Matthiessen, 1993). In the next section a conceptual framework of the impact of improving cross-border communications on crime is proposed.

2 For a general review of studies dealing with borders, see Van Houtum (2000), Leimgruber (1981, 1998, 1999) and Bucken-Knapp and Schack (2001). 3 For a comprehensive discussion on the meaning of borders and crime in the former Soviet Union, see Galeotti (1995) and for borders in the context of cyber crimes, see Speer (2000). 7

Table 1 Susceptibility of border regions to crime

Type Factor Description Location and Border’s regional This may impede or facilitate criminal activity. Field et al., geography position. Border type (e.g, (1991) illustrate how type of border adjacency influences the length, landscape, type of routes of criminal activities at the Mexican-American adjacency whether land, border. Similar examples are presented by Vagg (1992) and sea or bridge). Hajdnijak (2002). Societal Economic inequality and Wage differentials between countries, or significant structure and relative deprivation unemployment on one side of the border stimulates organisational between neighbouring marginalised groups to see illicit business as a way of differences countries; cultural survival (Galeotti, 1995, Schhoenhardt 1999, Ruyver et al., differences and gender 2001). Sometimes economic inequalities do not need to be inequality large between countries to stimulate criminal activities (Anderson, 1995:22). Gender inequality leads to traffic in women and prostitution (Ruyver et al., 2001, Di Nicola 2001). Weakened state apparatus Criminal cross-border networks may take advantage of weak and political and social or corrupt state apparatus. Examples include the former instability Soviet Union countries (Galeotti 1995, Osyka 2001, Osmonaliev 2002) and former Yugoslavia (Hajdnijak 2002). Political differences and mutual hostility between adjacent states lead to criminal opportunities, e.g, between Northern and Southern Ireland, (Bew and Gillespie 1993, Wilson 2003). Differences in taxation, Differences in taxation, tariffs or regulations on products in tariffs and regulations neighbouring countries encourages smuggling Persson (1999), Rikskriminalpolisen (1999) and (Krajewski 2001). Differences in laws and For an offender, differences in laws create different levels of law enforcement and lack risk. The perception of profit from crime may also vary if of harmonisation of punishment differs between countries (De Ruyver et al. criminal justice/ 1992). Differences in law enforcement regimes (including legislation cultural differences) make it difficult for cross-border police to co-operate on a daily basis. See Junninen and Aromaa (2000). Conditions The symbiosis of Interaction between typical cross-border crimes (e.g., for criminal traditional and cross- smuggling and drugs related offences) and other types pf activity border crime crimes, may affect the volume and characteristics of the latter at or near the border. This in turn, creates a particular criminogenic environment (Field et al., 1991). Money laundering is one example of such interaction (Osyka 2001). Offenders’ knowledge For international criminal networks, having people with and perception of the good local knowledge is crucial. Perceived distance border (geographical, linguistic, cultural) ‘shrinks’ as soon as offenders have the necessary knowledge about the other country’s system to minimise the risks of being caught.

3. Impact of improving cross-border communications on crime: a conceptual framework

We now consider how geographical patterns of offences might be expected to change with the opening of the Öresund bridge. The hypotheses put forward are based on 8 established criminological theories and on scenarios suggested by the Swedish Police Authority and Customs Service Office Rikskriminalpolisen (1999). We start by briefly providing some basic concepts of crime at the border.

It is important to distinguish between traditional and cross-border crimes and the interaction between them when assessing the impact of an improved transport link on crime patterns in a border context. According to Anderson (1995:14-15) traditional border crimes are crimes that take place in locations along border but do not involve transporting illegal goods or services across the border (e.g., selling drugs in border villages) but the levels of these crimes and/or the forms they take may be affected by the presence of the border. Cross-border crime involves transportation of illegal goods (or amount of goods) or services across the border of a national jurisdiction. They may involve one or several crossings and one or several frontiers (e.g., smuggling). Porter (1996) suggests that cross-border crime is characterised by “an offender crossing a police force boundary to perpetrate a crime or commit an offence, which requires the police to cross boundaries to investigate it” (p.3). There are also ‘organised crimes’ that are composed of a series of individual crime incidents or transactions and that may have a European, a supra-European or a global dimension (e.g., traffic of drugs).

There are four processes that characterise the impact of improving cross-border communication on offences (Figure 2). Instead of being distinct, these processes often overlap each other (e.g., processes 1 and 4). Due to the geographical position of the Öresund bridge in relation to the European continent, we emphasise underlying processes as they relate to Denmark and Sweden.

The first process refers to the increase in mobility in the region of motivated offenders. When ‘a new element’ is added to the transport system across a border region, offenders may reassess the potential for new activities across the border, taking into account risks and benefits. For most offenders, the perceived costs are high, their behaviour is unchanged, the bridge affects them very little. For this group, crime is a local activity that is committed in ‘familiar places’ (Brantingham and Brantingham, 1981) or at relatively short distances to their homes (White 1932, Porter 1996, Costello and Wiles 2001). For a minority of offenders, the new transport link does affect their perception of risk-benefit, and the border provides new opportunities. This relatively small group of individuals is far from being homogeneous, which makes it difficult to generalise about the reasons why they perceive the bridge as providing an opportunity. This group may consist of professional bank robbers, thieves, individuals who get involved in violence (e.g., weekend troublemakers, teenage gangs involved in vandalism) but also ‘wise tourists’. A ‘wise tourist’ is an individual who crosses the border using the bridge and has planned in advance what to do on the other side of the bridge (e.g., purchase of illegal goods or services, such as prostitution and smuggling of alcohol and/or cigarettes). The experience with Euro tunnel shows that even traditional 9

Causal factor Criminogenic condition Offence outcome

1 • Greater mobility of ••MoreMore offences offences 1 • Increase in the routine certain motivated • Increase in the routine committedcommitted by by non non-- activity spaces of offenders offenders activity spaces of offenders nationalsnationals and and shortshort term term visitors visitors Intervening factors •More people unfamiliar ••ConvergenceConvergence of of motivated motivated ••’Tourism’Tourism criminality’, criminality’, LocationLocation & & RegulatoryRegulatory 22 with an area; spaces with offendersoffenders and and victims victims assault,assault, pick pick pocketing, pocketing, geographygeography frameworkframework at at high population turnover togethertogether with with a alack lack capable capable shopliftingshoplifting the/closethe/close to to the the guardiansguardians BridgeBridge Societal structure border: e.g., the construction Societal structure border: e.g., the construction && organisational organisational routineroutine practices practices • Changes in land use; differencesdifferences ofof police police and and • Changes in land use; creation of new sites ••NewNew patterns patterns of of activities activities ••PredatoryPredatory crimes crimes customscustoms officials officials creation of new sites 33 (links and nodes) byby motivated motivated offenders offenders and and (e.g.,(e.g., vandalism vandalism in in public public ConditionsConditions for for (links and nodes) potentialpotential targets targets andand private private places) places) criminalcriminal activity activity

• Increase in smuggled ••EasierEasier movement movement of of • Increase in smuggled ••NarcoticsNarcotics and and other other 4 illegal goods and people, goodsgoods (including (including humans, humans, offences related to 4 illegal goods and people, cargo and drugs) in the offences related to e.g.,e.g., ’Myrtraffiken’ ’Myrtraffiken’ cargo and drugs) in the addictionaddiction and and smuggling smuggling borderborder region region

Figure 2 - The impact of improving cross-border communications on crime: a conceptual framework. 10 crimes may increase due to the inflow of offenders coming from the other side of the border but the effects may be asymmetric. ‘Criminals living in France do not travel to offend in the UK to the same extent as criminals living in the UK. The UK’s offenders use the tunnel to travel to France and on to Luxembourg where they commit robberies, especially in shops and petrol stations. There are cases also of robbery against cargo transport in Belgium being committed by UK criminals’ (Rikskriminalpolisen 1999). Therefore, an increase in offences committed by non-national offenders and short-term visitors is expected to take place in the region as a direct effect of bridge construction.

The second process refers to the fact the bridge facilitates the inflow of transient groups unfamiliar with the environment which leads to an increase in so-called tourism- criminality (Rikskriminalpolisen 1999). More people does not mean better surveillance, quite the opposite; it may also mean fewer capable guardians. New travellers, unfamiliar with the levels and types of risks in particular areas, may be targets if motivated offenders are drawn to these areas (Brantingham and Brantingham 1981). An increase in pick pocketing, shoplifting, theft, and in certain cases, violence, including hooliganism are to be expected. Central areas and their surroundings, where most of the attractions are located could become hot spots. The Öresund region was becoming a more integrated region in the 1990’s and the opening has accelerated this process. By 2001, Danes had virtually doubled their hotel nights per thousand in the Swedish Öresund region since 1997, whilst the figure has remained constant in the rest of the Sweden. In the opposite direction, from Swedish to Danish Öresund, the process has been similar (SCB 2003). Another indicator of the integration and greater mobility is the increased flow of people between Copenhagen-Malmö. The volume has increased 25 percent since 2000 (Figure 3).

30 000 000

25 000 000 s r e

g 20 000 000 n e

s Helsingör-Helsingborg s a p

15 000 000 Copenhagen-Malmö f o

r Öresund in total e b

m 10 000 000 u N

5 000 000

- 1998 1999 2000 2001

Figure 3 – Number of passengers in Öresund region, 1998-2001. Data source: Öresundsbro Konsortiet, 2002. *Öresunds bridge opened for traffic 1st July 2000.

The bridge is responsible for transporting the highest share both for train and private cars (Table 2). Figures from COWI (2002) show that people using the bridge have a very concentrated spatial mobility pattern around Öresund bridge. Swedish travellers travelling either by car or train over the bridge mainly had the Copenhagen area as their 11 destination (42 per cent of cars, 70 per cent of train passengers). It is worth noting that only 8 per cent of these had Copenhagen International airport as their final destination. On the Swedish side, as much as 36 percent of car trips had Malmö as their destination (78 per cent had Swedish Öresund as their final destination) whilst of those using the trains, 72 percent went to Malmö. Looking at a more permanent indicator of the region’s integration, such as the total number of inhabitants moving, it can be seen that whilst the total Danes moving from Denmark to Sweden increased more than 200 per cent, the opposite flow has remained the same since 1998 (SCB 2003). Differences in costs of living (food, housing), taxation and in the labour market between the two countries may explain this unbalanced flow.

Table 2 - Passengers and private cars over Öresund bridge (in thousands) Passengers 1999 2000* 2001 With ferry-boats 19,445 16,329 12,319 With train over Öresunds bridge 2,701 4,952 With car over Öresunds bridge 4,791 6,964 Total passengers 19,445 23,822 24,235

Private Cars With ferry-boats 2,588 2,188 1,851 Over Öresunds bridge 1,656 2,926 Total cars 2,588 3,844 4,777

Source: SCB/Öresundsbro Konsortiet/Malmö och Helsingborgs hamn, 2003. * Öresunds bridge oppned for traffic 1st July 2000.

The third process refers to changes in land use and hence crime in the region. The bridge and the whole transport and service related infrastructure that follows create new sites, new land uses that generate new patterns of activities and hence offences (Wikström 1991). Transport nodes of any kind are typical examples of poorly guarded places and increases in travel flows intensify the congestion and pressure on public transport system leading to physical deterioration in bus stations, railway stations and airports. An increase in offences is expected to take place in central areas (e.g., CBD and cultural quarters in inner cities) and at transport nodes, such as railway stations. Pick pocketing, all types of thefts, shoplifting and offences related to physical deterioration will increase. New sites associated with the bridge opening (e.g., tourist information centre, Cafés) and local train stations and Centralen/Skeppsbron in Malmö become particularly vulnerable places. Offences on buses, trains and boats, mostly theft and pick pocketing, are likely to be affected.

The fourth process deals with the direct impact of the bridge (and the new Customs Services routines) on cross-border crime and indirectly, on traditional crimes. It is expected that the bridge (especially through train services) will directly affect smugglers, especially drug users or small dealers that travel to Denmark with the objective of acquiring narcotics. The drug trade between Denmark and Sweden is a consequence of drugs in Denmark being cheaper, of better quality and easier to buy than in Sweden. In addition, there is a more liberal atmosphere towards drugs in Denmark (e.g., in 12

Christiania, a community close to Copenhagen city core that allows the selling and consumption of drugs). This activity involves the purchase of small quantities of substances, smuggling them to Sweden, either for consumption or selling. This is often called in Swedish ‘Myrtraffiken’ (Ants’ traffic) since it refers to the repeated smuggling of small quantities of narcotics by train. According to Rikskriminalpolisen (1999) the experience with the Euro tunnel shows an increase in narcotic crimes in the border region since France is a transit country to the UK - a track that starts in Morocco, passes through Spain and France and then to England. An increase in smuggling and narcotic related crimes is expected to increase after the bridge opens. Some indirect impact may be noticed also on other traditional offences that are related to drug addiction, such as residential burglary (Wiles and Costello 1999) and theft from cars. The literature shows that drug addicts may get involved in these types of offences in order to obtain money to buy drugs.

Figure 2 summarises the conceptual framework of this study. Intervening factors are factors that mediate the effect of the bridge on offences. The first group of factors are related to the nature of the border or of the countries where the border is located. The importance of each factor has already been discussed in section 2.2 and summarised in Table 1. The second group of factors relate to the characteristics of the border as a controlling organ of the national boundary. This includes the type of control processes employed at the border (e.g., sampling procedure) and police practices. Both of these are dependent on political agreements (e.g., Schengen agreement, Nordic countries agreement) and local organisation (staff size, organisation and co-operation with other forces).

The sampling procedure for border checking influences the chances of being caught (Ahlberg and Thedeen 1981). Empirical evidence (Tsebelis 1989, Sherman 1990) suggests that offenders adapt to changes in police practice and vice-versa, thus, changes in the control procedures at the border should directly affect offenders, and in turn, Custom service routines should also change as a result of changes in offence volume. For organised criminal leagues, the new checking system affects their perception of risks and benefits; in certain cases, a change may simply mean a change in costs. Costs are a function of the risks of detection at the border but also of structural factors that may vary between the bordering countries. These costs relate to punishment length and the likelihood of being caught, state intervention on prices of goods that may make the business (un) profitable, supply and demand in a regional context, the integrity of state and police institutions. There are also personal costs, such as the risk of physical harm, stigma and rejection, guilt and shame when caught.

Before the bridge, boats were the only means for crossing Öresund, and the Custom Services control operated in a different manner to today. In the past, part of the ‘selective control procedure’ on cars, lorries and people was done on the boat during the trip. Selective control procedures involve “a controller’s professional experience and intuition for sampling, checking of information and objects with certain risk profiles to the use of trained dogs”… (Rikskriminalpolisen 1999:115). Nowadays, with the bridge, the procedure is still ‘selective’ by taking samples of pay-and-go travellers at the border. It is 13

suggested that vehicles which pay electronically are underestimated in the sample since they hardly stop at the border. This may facilitate smuggling. In the case of the Euro tunnel (Rikskriminalpolisen 1999), for instance, there is a consensus among the police authorities that the ‘tunnel facilitates the flow of criminals between the UK and the continent and consequently, an increase in crime and the entrance of illegal migrants in comparison with ferryboats’. So, it is expected that the bridge will increase smuggling of goods such as alcohol and cigarettes and illegal immigration as well lead to an increase in traditional crimes as a result of the inflow of criminal organisations, perhaps from other Baltic countries.

Police practice (including Custom services) is also an intervening factor that influences the effect of the bridge on crime. This includes the levels of policing and also on the level of detection and reporting in the region. So far, only partial information on this is available, which does not allow systematic analysis. One example refers to adoption of the Schengen agreement by of Nordic countries (including Norway and Iceland) from March 2001. In theory, this is important because it changed border routines, facilitating the flow of people in and out the Nordic countries. The Schengen agreement allows countries to remove their internal borders and allow people to travel without checks from country to country. This means that at the zone’s external boundaries (the borders with countries that are not a party to the Schengen agreement, in the Swedish case, these include all eastern harbours) all travellers are to be checked in the Schengen Information System (SIS) at special checkpoints and during predetermined opening hours. It is believed that SIS as a tool at the borders allows better control through the co-operation of Custom Services, Police and Coast patrol. In practice, the Schengen agreement did not change much the routines at the borders for Nordic citizens because they had this right assured by previous regional agreements. The major difference involves the control of the Nordic frontiers in relation to non-national Schengen citizens coming from or through continental Schengen countries.

The Swedish Police together with the Swedish Custom Services suggested the expected consequences of the Öresund link on crime in the region in a 1999 report (Rikskriminalpolisen 1999). The impact assessment, to the year 2005, was based on criminological theory and professional experience of crime patterns at the border and in the largest cities of the region.

Table 3 Offences Affected by the Bridge – OAB Direct impact of the bridge Indirect impact of the bridge Smuggling Residential burglary (includes cellar, attic, summer house) Pick pocketing Narcotics Vandalism Assault/mistreatment (includes those resulting in death) Molestation Personal robbery Vehicle theft Shops - robbery, burglary/theft, shoplifting Theft from vehicles/burglary Crime against common order Robbery in banks and post offices Theft or burglary in churches, museums, sport arenas Others: Illegal migration, taxes evasion crimes, prostitution and gang-related crimes (e.g., bikers, such as Hell’s Angels and Bandidos) Source: Rikskriminalpolisen (1999), p. 86. 14

According to Rikskriminalpolisen (1999), the impact of the bridge would depend on the type of crime. It was not expected that all crimes would be affected. Table 3 differentiates between those which the bridge would have a direct impact on (e.g., smuggling and pick pocketing) from those which the bridge would have a long term or indirect impact (e.g., narcotics, personal robbery). This selection of crimes will be used in the analysis in section four to six. Offences, such as, prostitution, gang related crimes, illegal immigration were considered special cases. This latter category will not be considered further in this analysis. For a full description and crime codes used in this paper, see Appendix 3.

An underlying problem in subsequent analysis is to disentangle bridge effect from other contextual effects, such as local and national trends in crime but also with wider societal changes such as levels of criminality and changes over time of the perceived benefits from engaging in different criminal activities. The approach here will be to compare national data totals (Sweden and Denmark) in order to try to benchmark the regional data and provide a framework to separate regional from national trends.

4. Crime before and after the opening of the Öresund bridge

4.1 Regional crime levels in Öresund We first consider all crime types in the region focusing on the differences between Swedish and Danish Oresund. There is a more detailed consideration of human smuggling and driving under influence. This is followed by a discussion of those offences that were expected be affected by the bridge (OAB) for both the whole region and the two countries separately.

Thefts, shoplifting and vandalism are, by far, the most common types of offences in the Öresund region. However, there are differences between the Swedish and Danish sides. In Swedish Öresund the most frequent crimes in 1998 were theft from cars, theft of bicycles, shoplifting, thefts-others, car thefts, vandalism in vehicles, vandalism in general, pick pocketing, residential burglary (in cellars/attics) and threat, in that order. In the Danish side, thefts-others, theft of bicycles, theft from cars, shoplifting, possession of narcotics, car thefts, crime against common order, vandalism, residential burglary, burglary in shops are the top ten offences in 1998. The rank in 2001 for both Swedish and Danish Oresund is almost identical. These marginal shifts in rank may reflect more the opportunistic and hence substitutable nature of these offences (such as, car related thefts and shoplifting) than the bridge’s impact. Most offenders are flexible in engaging in different types of illegal activities depending on the circumstances.

However, not all shifts can be explained by situational conditions that may increase one type of offence compared to another. One example is the significant increase of human smuggling (trafficking) and/or illegal migration4. Both in Sweden and Denmark human 4 The terminology varies in the literature since they constitute a set of offences that are regarded differently from country to country. In Sweden, for instance, it may involve at least three different offences, from helping an individual to come and stay in the country illegally to organised schemes to bring foreigners to the country illegally for profit. 15 smuggling is a new phenomenon. In a European perspective, the so-called ‘Baltic Route’ has been increasingly exploited since the collapse of the Soviet Empire (Di Nicola 2001). In Swedish Öresund, there has been an increase from 2 cases in 2000 to 49 in 2002, 46 cases recorded in Malmö alone. These are indications that the bridge has facilitated human trafficking, a phenomenon that cannot be explained by local factors alone. As Schhoenhardt (1999) suggests, trafficking organisations exploit changes in border control systems, enabling them to change routes temporarily and take advantage of loopholes in border surveillance and controls. In Denmark figures from NCPD and Police (2002) show that women from the Baltic States (particularly Latvia) and Asia (particularly Thailand) were brought to Denmark and forced into prostitution. According to Schhoenhardt (1999), human trafficking has become a growing business because of the restrictions on legal immigration imposed by industrialised countries, the relatively low risks of detection and, the growing demand for international migration, mostly in sending countries but also to some extent, in receiving countries.

Table 4 – Changes in OAB (Offences Affected by the Bridge) totals 1998 to 2001. Swedish Observed Danish Observed Öresund Sweden -expected* Öresund Denmark -expected* OAB (%) (%) (Counts) (%) (%) (Counts) Smuggling ** ** ** -35,4 -44,3 14,0 Pocket picking -2,5 8,7 -489,1 5,5 7,9 -1107,7 Vandalism 8,2 15,1 -958,8 4,4 4,1 40,4 Molestation 27,2 20,6 157,6 35,5 22,6 92,4 Vehicle theft -2,1 -4,0 171,2 -23,8 -18,6 -890,1 Residential burglary -25,1 -21,6 -297,7 -12,8 -4,9 -1472,6 Narcotics 7,5 2,7 175,5 -19,5 -9,3 -854,4 Assault/mistreatment 10,6 4,5 407,8 11,5 18,0 -201,1 Personal robbery -2,2 8,3 -90,9 11,3 23,5 -113,9 Shop related offences -16,2 -18,3 186,0 -24,9 -20,4 -808,8 Theft from vehicles/burglary 6,5 -5,8 2620,3 -13,1 0,8 -2719,0 Crime against common order 35,6 25,0 38,6 32,5 27,2 34,3 Robbery in banks/post offices -25,0 20,4 -5,4 81,3 70,9 6,6 Theft or burglary in churches -23,0 7,5 -72,8 -19,6 -8,8 -147,8 Total OAB 0,0 -0,7 1423,4 -7,8 -2,4 -8127,8 Total crimes in Sweden & Denmark 0,7 -5,18 *Observed = total number of offences in the region; Expected = an estimate of the total number of offences in the region if the region had followed the national trend in crime levels between 1998 and 2001. ** See Table 7.

The category of offence that has increased most in Swedish Öresund is thefts of different types but mostly from cars and of bicycles. Traffic related offences (mostly driving under the influence of substances such as alcohol or narcotics) have also increased. However, such an increase may partially be related to changes in the way the offence is recorded5 or

5 Driving under influence of substances was recorded separately from other types of offences but lately there has been a common practice to record it in the same database as other offences. One reason is that driving under influence of alcohol in combination with narcotics has become very common during the last years – a fact that confirms the increase of narcotics in general but may have inflated the number of cases since the offence was possibly recorded in both databases. 16 may be linked to the increase in short weekend trips from Swedish Öresund, mostly Malmo to Copenhagen. According to Bygvrå and Westlund (2001) the purchase of beer and wine and dining in restaurants and bars were the main reasons Swedes travelled to Denmark (as was the case before the bridge). Another explanation is related to easier purchase of small quantities of illegal substances which are then smuggled across the border. In the Danish Öresund, figures for 2001 NCPD and Police (2002) show that, in comparison with previous years, there was a marked increase in the number of particularly dangerous robberies primarily targeting public buildings, petrol stations and valuables in transit.

Table 5 – Average percentage change 1998 to 2001 in OAB , other offences and total. OAB Others Total Swedish-Öresund 0,0 0,7 0,3 Danish-Öresund -7,8 -7,2 -7,7 Sweden -0,7 2,5 0,7 Denmark -2,4 -11,9 -5,2

Tables 4 and 5 summarize changes in offence totals. The OAB-category levels in Swedish Öresund have not changed since 1998. However, the region had in 2001 around 1423 offences more than it would have had if the region had followed the national trend. In Denmark, the region had 8128 fewer offences than it would have had if the region had followed the national trend.

Among the OAB-category, vandalism, molestation, assault/mistreatment and crime against common order increased between 1998 and 2001 in the whole Öresund region. In the case of molestation and crime against common order the increases exceeded the national trends in both countries. These offences are typically concentrated in large agglomerations of people and at sites of tourism and leisure activities. As noted earlier, the bridge increases mobility and interaction within the region (see Figure 2 and the discussion of the second process) creating new criminogenic conditions that did not exist before the bridge. From a victim perspective, more people now are exposed to unknown environments and situations making them 'targets’, vulnerable to crime. From an offender point of view, new sites and new targets (visitors, tourists coming to their areas of activity) create new opportunities for crime.

There are differences between Swedish and Danish Öresund in shifts in OAB. This may reflect the asymmetric mobility pattern between the regions. In the Swedish Öresund, vandalism, molestation, narcotics, assault/mistreatment, theft from vehicles and crime against common order increased between 1998 and 2001. Figure 4 illustrates the increase in two offences that occurred at more or less the same time as the bridge was opened on 1st July 2000. By comparing the graphs of year 2000 with the ones of 1998 and 2001, we confirmed that such variation does not follow only seasonal patterns of variation. For narcotics in 1998, only January had around 200 cases whilst other months had 130 cases on average. In 2001, the highest levels were around 200 cases but with an average of 180, which is lower than 2000 but much higher than 1998. A similar but less distinct pattern was found for car related thefts as well. 17 Narcotics 2000 Car related offences 2000 Bridge opening July 2000 Swedish Öresund 300 3200

2800 200 2600

2200 100 2000 t t

n 1800 n u u o 0 o

C 1600 C 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 Bridge opened Bridge opened MONTH MONTH

(a) (b) Figure 4 – Increase in narcotics (a) and car related thefts (b) in Swedish Öresund in 2000. The lack of time information varies slightly from year to year. As much as 20 per cent of offences did not have any data on month in 2000.

In Denmark’s Öresund region, increases are recorded in pick pocketing, vandalism, molestation, assault/mistreatment, robbery in banks and post offices and crime against common order. For all OAB-category, the Copenhagen metropolitan area concentrates the largest share. Although the Swedish Öresund has witnessed a slight increase for both OAB and other type of offences, Malmö (the Swedish city border to which the bridge is linked) has recorded a decrease in the number of offences between 1998 and 2001 (Table 6).

Table 6 - Percentage changes in OAB totals 1998 to 2001:main urban areas in Swedish Öresund OAB Others Malmö -7,4 -2,2 Swedish Öresund (excluding Malmö) 3,9 2,1 Helsingborg -4,9 -6,3 Lund 8,4 3,2 Kristianstad -10,1 1,3 Ystad 2,3 13,2 Swedish Öresund 0,0 0,7 Sweden -0,7 2,5

Other cities close to the Swedish border have also shown an increase in total crimes regardless of type; Lund and Ystad show such an increase. In other words, the hypothesis that areas close to the bridge would have higher increases in OAB offences than the areas far from the border cannot be corroborated by these findings. One possible explanation for this may be changes in Police enforcement routines leading to more pro-active crime prevention measures since the bridge opened. Another reason may be related to 18 adaptations of Custom Services routines to the so-called “unknown threats” for certain types of cross-border crimes (Rikskriminalpolisen 1999).

4.2 Cross-border crime levels in Öresund: the case of smuggling Differences in taxation between Sweden and Denmark and the Baltic Countries in products such as alcohol and cigarettes stimulate smuggling. Beer in Denmark is three times cheaper than in Sweden and such differences in price stimulate Swedes to bring these goods illegally into the country. According to Bygvrå and Westlund (2001) Swedes increased not only the volume of trips to Denmark after the opening of the bridge, but also spent more on arrival. Swedes in 2001 spent, 25 per cent more than in May 2000, mostly on beer and wine. Part of this is almost certainly brought into Sweden by weekend smugglers for domestic purposes. According to Rikskriminalpolisen (1999) since 1998, there were also cases of well-organised smuggling leagues caught with large quantities of goods. Cigarette smuggling has also increased between 1998 and 2001. According to Persson (1999) it is worth noting that 3 per cent of arrests in a year are responsible for around 90 per cent of the total seized cigarettes in a year, often driven by organised smuggling gangs. However, most of the arrests consist of ‘domestic users’ (husbehovsmugglarna) bringing in small but illegal quantities. The gangs by contrast are internationally organised groups that have their wholesale dealers in Northern Europe. The Baltic region states plus the Netherlands are the source states for this type of smuggling. Offenders are often versatile and smuggle combinations of goods: cigarettes, alcohol and narcotics. In mid-1998, a tax reduction on cigarettes was implemented in Sweden to try to contain the increasing illegal market, however, according to Rikskriminalpolisen (1999) cigarette smuggling has not been affected by this intervention.

The increased movement between Sweden and Denmark following the bridge’s opening provides ample opportunity for smuggling. From and through Denmark into Sweden, smuggling involves mostly alcohol, cigarettes and narcotics. In absolute numbers, the total number of seized goods at Customs Services in Swedish Öresund has increased since 1998, especially in Malmö (Table 7). These findings go in the same direction as the ones discussed by Di Nicola (2001) for cross-border crime between Eastern Europe and the European Union.

However, if we compare the total number of seized goods with total number of passengers or with the total vehicles crossing the region, then the rate is unchanged or, has even decreased since 1998. Improvements in control procedures may have increased the Custom Services ability to catch smugglers. To the extent the rate has gone down it is either due to better customs/tracking procedures or because the bridge has increased the proportion of people who do not intend to smuggle goods or it could be that the volume is so huge that the risk of detection has actually fallen. Table 7 suggests a clear displacement from Helsingborg to Malmö as a result of the Öresund bridge, indicating that smuggling has changed its geography. Smugglers seem to prefer the bridge route between Copenhagen-Malmö rather than the ferryboat between Helsingor-Helsingborg. According to Customs, control routines have not changed significantly since the bridge was opened. This lends strength to the argument that there has been a change in the 19

geography of smuggling. For instance, the number of personnel employed for control purposes, which could affect the number of seized goods, remained the same in Helsingborg and has only marginally increased in Malmö. However, according to Rikskriminalpolisen (1999) border controls are flexible and may require short-term adjustments, based on the anticipated volume of possible threats.

Table 7 Number of goods seized by Customs in 1998 and 2001 (by type and total number of passengers and vehicles) 1998 2001 Malmö* Helsingborg** Region Sweden Malmö* Helsingborg** Region Sweden Seized goods 1571 2330 3901 16301 3728 1376 5104 14415 Narcotics 509 456 965 2065 1590 535 2125 3547 Fire weapon 4 1 5 60 19 5 24 1303 Tobacco 423 317 740 6859 843 40 883 3483 Others 635 1556 2191 7317 1276 796 2072 6082 Seized goods 0,30133 0,17056 0,20668 - 0,26677 0,00012 0,20007 - /passengers (Thousands) Seized goods 4,46204 0,88658 1,30899 - 1,26335 0,60439 0,97656 - /vehicles (Thousands) *Copenhagen-Malmö – up to 1999, ferryboats carried out the service for passengers and vehicles. In 2000, the Öresund bridge takes over this flow. **Helsingor-Helsinborg – only ferryboats Data source: Smuggling: Swedish Custom Services, 2002; Passengers/Vehicles:Öresundsbro Konsortiet, 2002.

The total number of narcotic related substances seized by the Swedish custom services in Malmö increased from 509 to 1590 between 1998 and 2001. Narcotics include: cannabis (from Morocco-Spain, the Netherlands and Denmark), heroin (from Afghanistan, Pakistan-Balkans, Eastern European countries-EU), amphetamine (from Benelux, Poland and Czech republic), cocaine (from South America) Rophynol and ecstasy (from Spain, Greece and Central European countries). Private cars or lorries are mostly used for smuggling cannabis, amphetamines and heroin. Smuggling cocaine and doping substances is mostly carried out by commercial flights or by mail.

4.3 Offender mobility after the opening of the bridge It is difficult to establish the extent to which the new bridge has led offenders to cross the bridge to commit offences. Taking Swedish Öresund as an example, we checked whether or not offenders were more willing to cross the border after the bridge opened in 2000. The available data was on total number of suspects committing offences by citizenship and Police authority district (Table 8).

Despite the fact Table 8 does not provide an accurate indication of offenders’ mobility pattern since we do not know their home address, this data constitutes ‘a good proxy’(Rikskriminalpolisen 1999). As much as 75 per cent of Polish citizens and 45 percent of Danes arrested for committing crime in Swedish Öresund in 1998 did not live in the region. Among Polish citizens, a large number come from deprived areas in 20

Northwest Poland. Although there is recent evidence of Polish organised crime leagues in Swedish Öresund, Polish offenders are often people who engage in shoplifting and smuggling of narcotics, or people (age 40+) who travel to Sweden to sell illegal goods. Danish offenders were mostly young offenders who engage in shoplifting, vandalism, fraud and smuggling of cigarettes and narcotics. Among those who were on trial in Sweden, the majority was involved in traffic related crimes (e.g, drunk driving), narcotics and smuggling. None of this may be affected by the building of the bridge. In the case of Danish drug sellers, for them, ‘there is no need to cross the border to reach the Swedish market since clients will come to them” (RKP, 1999:42).

Table 8 - Suspects in Swedish Öresund*, by selected citizenships Danes Germans Poles Latvians Lithuanians Estonians Total M Öre M Öre M Öre M Ore M Öre M Öre 1998 485 1621 80 290 755 2257 12 35 26 235 16 63 4501 1999 497 1623 86 291 723 2254 9 27 30 222 16 50 4467 2000 533 1716 57 253 808 2252 11 26 40 251 19 54 4552 2001 469 1661 84 260 788 2271 9 28 27 244 9 45 4509 2002 544 1653 98 276 806 2302 2 26 22 241 14 56 4554 Data source: RAR database, Skåne Police Authority, 2003. * M = Malmö Police Authority area, the closest police district to the Öresund bridge, which has the largest crime frequency in the Öresund area. Öre = Swedish Öresund.

As Table 8 shows, there has been a slight increase in the total number of offenders in Öresund for all selected citizenships, with the exception of Germans and Latvians. However, in the Malmö Police authority area, there has been a decrease in the number of suspects from all Baltic countries, with the only exception being Poland. For geographical reasons, the ferryboats are still the easiest connections from Baltic countries to Southern Sweden. In long term, new restrictive controls at harbours on the Swedish east coast may force offenders to choose the bridge as the alternative for smuggling, following the trend of increased human trafficking.

5. Changes in the geography of offences: empirical results of Swedish Öresund

In this section we examine changes in the geography of offences that might be a consequence of the bridge opening, using the Swedish border municipality of Malmö as a case study (Figure 5). Although crime levels have not increased in Malmö during the studied period, it is expected that the geography of offences could have been affected by changes in people’s routine activity and criminogenic conditions. An increase in offences was expected to take place in the surroundings of central areas (CBD, commercial and cultural inner cities) and in transport nodes, such as railway stations. The increased mobility would generate new clusters of so-called tourism criminality (pick pocketing, all types of thefts, shoplifting) in the inner city areas. Local train stations and places like Centralen/Skeppsbron in Malmö become highly vulnerable places also for vandalism. An increase in smuggling of narcotics should impact not only on the geography of narcotic- related offences in the central areas but also in some traditional crimes all over Malmö, such as car related thefts and residential burglary. 21

The methodology involved first the creation of a spatial framework, which merged spatial units with small populations into larger zones; second, the calculation of changes and standardised offence rates on this spatial framework and finally, the detection of clusters of offences before and after the bridge was built. A symmetric time interval was used to make this analysis, that is to say, 18 months before and 18 months after the bridge (1999 and 2001). Robbery in banks and post offices and theft/burglary in churches, museums, etc were not analysed in this section since there were too few events.

Figure 5 – Main landmarks in Malmö, the Swedish border municipality in Öresund. 22

In order to create robust units of analysis, the initial 400 zones over Malmö were aggregated using the zone design software Z-DES6 (Openshaw et al., 1998). Aggregating small regions or merging them into larger ones produces area rates that are more robust and have smaller standard errors (Haining 2003). Of the three criteria available in Z-DES, population homogeneity was used combined with a shape-constraint function. The homogeneity criterion ensures that the population in the new spatial units is as similar as possible whilst the shape-constraint function

produces polygons that are similar in shape to the original ones. This is an advantage but there is a risk that the minimum population constraint may be violated when using the shape-constraint function. From a set of maps containing 50, 60, 70,80 and 90 new units obtained from Z-DES a map with 70 zones was chosen. This map was chosen based on a subjective assessment of what the minimum population ought to be to try to ensure reliable rates. Only one zone violated the population minimum so it was manually aggregated to a neighbouring polygon. The final number of zones was 69 from the original 400. The final set of zones was created in which the minimum population size was 998, the maximum was 7836, the mean was 3748 and the standard deviation was 1617. The final stage of the process involved the creation of a new set of boundaries by removing the boundaries between zones in the same region and merging their values together using GIS. Figure 6 illustrates the final set of regions containing the 69 new spatial units.

Figure 6(a) Changes in selected OAB in Malmö (%), 18 months before and after the bridge; (b) Areas with an increase for one or more offences.

6 http://www.ncl.ac.uk/sarg/zonedesign/

The count of six out of the eleven OAB increased slightly during this time period in Malmö but when aggregated all eleven OAB the count remains almost the same between 1999 and 2001. Figure 6 illustrates the changes for thefts from vehicles and narcotics (a) as well as the areas with an increase for one or more offences (b). Car related thefts (both theft from vehicles and vehicle thefts) have the highest increase in areas surrounding the Station Malmö Syd Svågertorp, possibly as a result of the increase of parked cars in this area. Shop related thefts had the highest increase in the harbor and central areas, mostly in the Northeast where a big retailer is located. The increase in residential burglary was concentrated in the southern part of Malmö, such as in Fosie, one of the most ethnically segregated residential areas (Bevelander et al, 1997). The geography of increases is similar for robbery, assault and narcotics, with the inner city and the parts of Hyllie having the highest increases. These findings suggest that the increase in narcotics, a typical cross-border related offence, has affected the dynamics of violent offences, such as robbery and assault.

A standardised offence ratio (SOR) for all selected offences was also calculated. This includes both those offences where it was hypothesized the bridge might have an effect (OAB) and other offences (‘controls’). This type of standardisation is a useful way of representing data for a set of areas where the areas differ in size (absolute values would tend to over-emphasise large areal units) and where it is necessary to allow for differences in population characteristics between areas (Haining 2003). The SOR for region i is given by (1): SOR(i) = [O(i) / E(i)] x 100 (1) Where O(i) is the observed number of offences of a given type and E(i) is the expected number of offences of a given type. The number of offences per individual at risk was obtained by dividing the total number of offences in the study area (Malmö) by the total size of the chosen denominator. For all cases, the chosen denominator was total population (for further discussion of this issue see Wikström (1991). For each area i, this average rate is multiplied by the size of the chosen denominator in area i to yield E(i). E(i) can be interpreted as the expected number of cases of the particular offence in area i under the assumption of a random distribution of offences across the whole study region. It provides a baseline from which to analyse the variation in O(i) since the expected value of O(i) is represented as the product of E(i) and the quantity r(i) which is called the area specific effect or relative risk (see, for example, Haining 2003).

The standard error for SOR(i) is given by SE(i) =O(i)1/2/E(i) and a 95 per cent confidence interval is obtained by multiplying SE(i) by 1.96 and this quantity is added and subtracted to SOR(i). A significant difference between SOR(i) and SOR(j), for example, at the 5 per cent significance level would be indicated if their confidence intervals did not overlap. Preliminary comparisons of SORs for the 69 spatial units showed there was no significant change in their geography before and after the bridge for any type of offence (This does not mean there was no significant change in the volume of the offences in each spatial unit). 24

A cluster detection technique was applied to investigate local changes in the pattern of SORs. In order to detect changes in the geographical cluster of offences, Kuldorff’s scan test was used (SaTScan version 2.17, (Kuldorff et al. 1998)). This technique has a rigorous inference theory for identifying statistically significant clusters (Haining and Cliff 2003). The test uses the Poisson version of the scan test since under the null hypothesis of a random distribution of offences (with no area specific effects) the number of events in any area is Poisson distributed. This test adjusts for heterogeneity in the background population. Because of the non-longitudinal data and because we are interested in comparing the situation before and after the bridge the spatial scan statistic was used on the two time periods separately.

Table 9 Most-likely and secondary clusters before and after the Öresund bridge: Malmö.

* Type Offences 18 months before (1999) 18 months after (2001) at No. significant* No. significant No. significant No. significant High value Low value High value Low value Clusters/ Clusters/ Clusters/ Clusters/ No. Polygons No. Polygons No. Polygons No. Polygons OAB Pick pocketing 1/5 0 1/5 0 Vandalism 1/6 1/30 1/13 1/29 Direct Molestation 0 0 1/5 1/29 impact Vehicle theft 1/6 1/30 2/6;5 2/29 OAB Residential 0 1/29 0 1/29 burglary Indirect Narcotics 1/13 1/30 1/5 0 impact Assault 2/6;11 1/29 1/5 1/29 Personal robbery 1/6 0 1/6 1/16 Shops related 1/5 0 1/5 1/16 offences Theft from 1/6 1/29 2/6;5 1/29 vehicles Crime against 1/6 1/30 1/9 1/29 common order Control Domestic 1/5 1/17 1/32 1/24 violence Fraud 1/5 1/20 1/6 1/21 90 per cent or 95 per cent. Note: 2/6;5 means two clusters, the most likely cluster is composed of 6 polygons and the secondary one is composed of 5 polygons.

The spatial scan statistic imposes a circular window on the map, which is in turn centred on each of several possible centroids positioned throughout the study area. For each centroid, the radius of the window varies continuously in size up to a maximum window size that includes 50 per cent of the spatial units. The circular window is flexible both in location and size and is moved across the map to search out the most significant clusters irrespective of size. The spatial scan statistic uses a large number of distinct geographical

7 http://dcp.nci.nih.gov/BB/SaTScan.html 25 circles, with different sets of neighbouring polygons within them, each being a possible candidate for a cluster. Each spatial unit is represented by a centroid, which determines for any given window whether the spatial unit is inside or outside the window. The likelihood function is maximized over all windows, identifying the window that constitutes the most likely cluster (that is to say, the cluster that is least likely to have occurred by chance). Its distribution under the null-hypothesis and its corresponding p- value is obtained by repeating the same analytic exercise on a large number of replications of the data set generated under the null hypothesis, in a Monte Carlo simulation. For more detailed information on the spatial scan test, see Kuldorff (1997) and Hjalmars et al., (1998). Table 9 reports for selected offences in Malmo the number of significant clusters, the most likely and the secondary ones. This is undertaken for high and low values offences (hot and cold spots). The number of polygons in each cluster is reported. There have been some small shifts in the geography of the selected offences, notably in relation to the location and size of the hot spots. The geographical pattern of seven offences out of eleven has changed, becoming either larger or smaller than it was 18 months before the bridge (Table 9, Figure 7). However, it is very difficult to argue that these shifts are due to the effects of the bridge. There are two reasons for this conclusion. First those were shifts not only in the selected OAB offences but also in the control offences. This indicates that variation in the geography of offences may be related to changes in other intra-urban criminogenic conditions (e.g., a rise in unemployment in certain areas, housing turnover). Second, no particular cluster was found in the immediate vicinity of the terminal of the bridge itself or in the surroundings of the secondary train station (Malmö Syd station).

Among the offences that the new bridge was expected to have a short-term impact, only the pick pocketing cluster is unchanged. This means that any change of the rate of this offence was not strong enough to affect its geography. For car related thefts, the most- likely cluster was not only larger in 2001 but a secondary cluster appeared. The first cluster includes the Kirseberg district (Figure 5), particularly the recreational and industrial areas and in the vicinities of a large retailer. After the bridge was built, the surroundings of the central station, harbour and surroundings became significant. It is possible that the bridge indirectly increased the car stock in these areas and, consequently, the number of targets for possible vehicle thefts and thefts from vehicles. Travellers can now easily go by train to the continent and therefore, cars need to be parked close to the station and its surrounding area.

The vandalism cluster became larger. For the offence of molestation the only significant hot spot appeared after the bridge had been built, in both cases in the inner city area, including central station, old town, eastern parts of the harbour and Kirsberg/Husie’s recreational areas (Figure 5). Only clusters of narcotics and assault have become smaller and more concentrated geographically in the city centre, railway station and harbour areas. Contrary to what was expected, no changes were found in the location of clusters for shop related offences (shoplifting, robbery or burglary) and personal robbery; both are also very concentrated in the inner city areas. The most surprising shift was in the geography of assault/mistreatment. Whilst before the bridge clusters were located in 26

Southern and Eastern parts of the city, after the bridge, the most-likely cluster had moved to Northern parts of Malmö. It is not clear why this should have happened. 27

Before the bridge After the bridge

Figure 7– Most-likely (dark shading) and secondary clusters (light shading) in Malmö: (a) vandalism (b) assault and (c) theft from vehicles, before and after the Öresund bridge.

6. Modelling changes in the geography of offence patterns In this section we model the geography of offence patterns before and after the bridge using demographic, socio-economic and land use variables covariates. The purpose is to explain the variation in area specific or relative risk (r(i)) for various offences in Malmö. 28

The dependent variables in this study are the standardised offence ratios for the eleven selected offences that were expected to be impacted by the bridge. Two other offences were used as controls (domestic violence against women and children and fraud in hotels, restaurants, fraud against disabled people and fraud-other). The control variables were chosen because they were high volume and were not expected to be affected by the bridge’s construction. Analysing these offences provides a check on the regression modelling in relation to the OAB-category.

We assess the impact of the bridge in two ways: (1) by fitting a single model that contrasts 1998 and 2001 data (2) by including in the model a measure of distance of each spatial unit from the bridge (through the variable Acess). The model controls for built form and land use features of Malmö and for population characteristics. Variables were chosen drawing on the existing urban criminology literature on social disorganisation, routine activities and the role of social capital (For a extensive discussion of these theories, see Entorf and Spengler (2002)). The choice of variables was based on the hypothesis that the bridge and regional structural changes as a result of its construction, may generate new patterns of offences by creating new sites for offending, stimulating offenders’ mobility and creating new groups of travellers that are more vulnerable to becoming victims of crime. To evaluate the bridge’s effect we must at least control for those variables that characterise criminogenic differences within Malmö (e.g., demographic and socio-economic variables) that are known to impact on offence patterns. Because of data limitations, the values of explanatory variables although from one time period, are not from the same year. The statistics for households are from the beginning of the 1990's whilst the statistics for local leisure associations are for 2002. For a further description of the variables, see appendix 3.

The linear regression model was used in order to explain the variation in offence ratios. This model is given by:

Y = X + ε (2) where Y is the vector of standardised offence ratios for 1999 and 2001 (18 months before and 18 months after the bridge was built) for a particular offence (N=69x2). X is an Nx(p+1) matrix with p+1 explanatory or predictor variables, including the constant term.

 is a p+1 vector of regression coefficients (0, 1,…,p) where 0 is the (intercept) parameter associated with the constant term and ε is the N vector of random errors with mean 0 and variance 2I. A dummy variable D was incorporated into the model. This is to assess the effect of the bridge, both on the intercept term and on each of the regression coefficients, so:

D(i) = 1 if the observation is from 1999, that is before the bridge was built (i =1,…,N) = 0 if the observation is from 2001, that is after the bridge was built (i =1,…,N)

In practice the first 69 values of the dummy vector consist of 1s and the second 69 consist of 0s. The full model is given by 29

T T Y = 01 + 0;DD +1X1 + 1;DD X1 + …+ pXp + p;DD Xp + ε (3)

Where DT denotes the transpose of the vector D and 1 is the N vector of 1s. The intercept coefficient is given by 0 after the bridge and (0+1;D) before the bridge. The regression th coefficient for the k covariate is given by k after the bridge and (k+k;D) before the bridge. We test for statistically significant differences before and after the building of the bridge by testing the significance of 0;D, 1;D, …, p;D from 0.

The regression analysis was implemented in SpaceStat 1.91 (Anselin 1992) since the software has regression modelling capabilities with a range of diagnostics (including heteroskedasticity tests) that are appropriate for spatial analysis (See also Ma et al., 1997). In order to test for spatial autocorrelation on the residuals, a row standardised binary weight matrix (W) was used that comprised non zero entries where i and j refer to areas that are adjacent and the SOR data refer to the same time period. The full weight matrix can be represented in partitioned form as follows: 1999 2001 W 0 Λ = 0 W

For five offences the OLS regression model diagnostics revealed significant spatial autocorrelation in the residuals. These are vandalism, vehicle theft, residential burglary, assault and shop related offences. Spatial lag and spatial errors models were fitted in order to address these problems (see Haining, 2003 for a discussion of when these different models are appropriate to consider). The matrix W was used to specify these models, which take the form of a lag operation on the response variable (spatial lag model) and the form of spatial correlation in the errors term (spatial error model). Table 10 summarises findings including the significant variables at the 1 per cent, 5 per cent and 10 per cent levels. After extensive analysis, heteroskedasticity remained in the residuals of most of the models and this is an area for further research. Not having a constant errors variance means that the regression estimates are no longer efficient. In the particular case of Malmö this problem is problem a consequence of using data from irregular spatial units (with different areas). For an extensive discussion of this diagnostic, see Anselin (1992), Haining (2003).

Only vehicle related crimes (vehicle theft and theft from vehicles) out of the eleven OAB-category have regression coefficients that are significantly different before and after the bridge’s construction. Model results indicate that the regression coefficient for central and local stations (and their surrounding areas), are significantly different before and after the bridge (DStat and Stat). We note that vehicle related crimes increased after the bridge opened. An explanation for this might be that the use of the bridge as a means to reach the continent (either by train or car) results in an increase in traffic flow in Malmö in general but especially close to transport nodes, such as stations. It may also indicate an increase in the stock of cars parked around the stations and surrounding areas that become vulnerable targets for theft and burglary. However, there are other variables 30 that help explain vehicle related crimes. Vehicle thefts, for instance, are concentrated in areas with a large share of population aged 25 and younger, high housing mobility, low foreign population and, surprisingly, with many local leisure associations. In the case of Malmö, local leisure associations, especially those related to sport events, may attract young motivated offenders who hang around and ‘blend in’ with the local environment. Thus, the presence of local leisure associations show little buffering effect on theft related crimes, which corroborates the findings of Ross and Jang (2000) but goes against results in Rosenfeld et al., (2001) and Martin (2002) on the impact of local organisations on crime.

Table 10 – Results of the regression analysis: Y = Standardised Offence Rates Type Offence R2 Significant variables Unsolved * = at 10 per cent, ** = at 5 per cent, *** = at 1 problems per cent OAB Pick pocketing 32.0 Urb**(+), Univ***(+), Shop*(+), Pbus(+) Heteroskedasticity Vandalism 42.2 W_Van***(+), Urb***(+), Shop*(+), Pbus(-) Heteroskedasticity Direct Molestation 26.4 Pfore***(-), Punemp**(-), Univ***(+), - impact Hosp***(-), Shop***(+) Vehicle theft 44.1 W_Veh**(+), Pout **(-), Pfore ***(-), Heteroskedasticity P25***(+), Plass ***(+), DStat **(-), Stat ***(+) OAB Residential 43.5 W_Res *(+), Pfore ***(-), Pvill ***(+) Heteroskedasticity burglary Indirect Narcotics 25.0 Pfore **(-), Ppfo **(+), Univ **(+), Stat ***(+) Heteroskedasticity impact Assault 28.9 Pfore***(-), P25***(+), Univ*(+), Heteroskedasticity Lambda ***(+) Personal 24.4 Pfore **(-), Ppfo ** (+), Access *(-), Univ - robbery ***(+), Shop ***(+) Shops related 15.9 W_Shop ***(+), Pfore ** (-), Univ ***(+) Heteroskedasticity offences at 90 per cent Theft from 40.0 W_Tve**(+), Pin**(+), Pfore ***(-), Plass Heteroskedasticity vehicles **(+), Dstat *(-), Stat ***(+) Crime against 25.9 Punemp **(-), Univ ***(+), Shop **(+) Heteroskedasticity common order Control Domestic 38.5 Pin***(+), DNws **(-),P25 ***(+), DUniv *(+) Heteroskedasticity violence Fraud 42.5 Pout ***(+), Univ***(+),Centre ***(-) Heteroskedasticity Access = Inverse distance to the bridge; Centre = Dummy for central areas; Hosp = Dummy for hospital and surroundings (1st order neighbours); Nws = Proportion of Neighbourhood Watch Schemes; P25 = Proportion of population aged 25 years and younger; Pbus = Proportion of buss stops per population; Pfore = Proportion of population born abroad; Pin = Proportion of population moving into the area; Plass = Proportion of Local Leisure Associations; Ppfo = Proportion of population with parents born abroad; Pin = Proportion of population moving into the area; Pout = Proportion of population moving out; Pvill = Proportion of single family houses (detached, semi-detached); Punemp = Proportion of unemployed labour force; Shop = Dummy for commercial areas and surroundings (1st order neighbours); Stat = Dummy for railway stations and surroundings (1st order neighbours); Univ = Dummy for university Campus and surroundings (1st order neighbours); Urb = Dummy for urban and urbanised areas. D_ represents the product of the time dummy (D) with the corresponding variable (e.g., Dstat). In the Space Lag model W_variable denotes the Lagged independent variable whilst Lambda belongs to the Space Error model. W_Van is the lagged dependent variable of Vandalism, W_Veh is the lagged dependent variable of Vehicle theft and W_Res is the lagged dependent variable of Residential burglary.

The presence of commercial areas (Shop) explains three out of the four OAB where the effect is expected to be direct (Table 10). Most pick pocketing occurs in the city’s main core, in commercial areas, close to the university campus and surroundings, and areas with relatively large numbers of bus stops. As hypothesized earlier, bus stops as transport nodes function as a meeting place for ‘obligatory outdoor activities’ (Gehl 1987). They 31 bring people together, creating an opportunity for crime (potential offenders and victims), which otherwise would be absent (Tremblay and Tremblay 1998).

However, whilst these places stimulate pick pocketing, they might discourage other types of offences, such as vandalism. High standardised vandalism rates are found in central areas, mostly commercial ones but in less guarded places, possibly with fewer bus stops. Areas with a high risk of molestation are associated with shopping areas. Population characteristics of the areas are also important - areas with a low proportion of unemployed people and population born abroad have higher vandalism ratios.

Among those offences selected where the bridge was expected to have an indirect or long term impact, the presence of the university campus and its surroundings (Univ) explains five out of these seven offences. As the variable is represented (the polygon and its first order neighbours), it signals the importance of educational environments on crime patterns. Other central areas, old town and the harbour area of Malmö are also important sites. A typical example is personal robbery. The model suggests that the closer an area is to a shopping area, the university campus and harbour areas, the higher the relative risk of robbery. Another example is narcotics. Areas with a high relative risk of narcotic related offences are the ones close to the stations (central and secondary ones), the university campus. As in the case of personal robbery, population characteristics are also important in explaining the geography of relative risk.

It should be noted that in the case of one control variable (domestic violence) there was a significant difference in two of the variables before and after the bridge. The relative risk of domestic violence was higher 18 months before the bridge was constructed close to the harbour/university campus and lower in areas with Neighbourhood watch schemes. This is disappointing in terms of the overall analysis since it undermines the regression findings on the OAB-category. It is possible, but unlikely, that the construction of the bridge may have had an effect on domestic violence but any link is likely to be tenuous.

7. Conclusions

This paper investigated the short-term effects on offences of the establishment of an improved transportation link between two bordering European countries. Results reported here suggest that new improved transportation systems that cross ‘open’ borders generates an increase in mobility, but this does not necessarily trigger a rise in offences. However the easy access to places may generate new patterns of offences by creating new sites for offending in areas that despite being crowded places, may lack capable guardians. New transient groups, unfamiliar with the levels and types of risks in particular areas, are targets for crime. In other cases, these new groups are the ones that become involved in opportunistic offences or other types of offences. An increase in travel flows and pressure on public transport systems, tends to result in an increase in vandalism and public order related crimes near stations, transit lines and vehicles. Moreover, a new transport infrastructure often leads to other types of investments that generate changes in land use. More parking spaces means an increase in the car stock thereby changing the areas susceptibility to vehicle related crime. 32

Thefts, shoplifting and vandalism are, by far, the most common types of offences in the Öresund region. It is suggested that there have been no dramatic changes in the total crimes between 1998 and 2001. However, changes in the transport system in the region have led to changes in smuggling routes. Smugglers seem to prefer the Öresund bridge route between Copenhagen-Malmö than the ferryboat between Helsingör-Helsingborg. There are also indications that the border control post bridge has facilitated human trafficking. As Sweden became part of the Schengen countries, the control over eastern coastal borders, regarded as external borders, became more restrictive, which means that is more difficult now to smuggle in people through these harbours. New routes of smuggling across the Öresund bridge became then the only alternative.

Results from the Swedish city of Malmö illustrate the fact that there have been some shifts in the geography of the selected offences, notably in relation to the location and size of clusters of high rates. The modelling of the changes in crime patterns show that only a few minor shifts in their association with underlying demographic, socio-economic and land use characteristics took place before and after the bridge was built. Only vehicle related crimes out of the eleven selected crimes seemed to be affected by the bridge. Space-time variation in one of the two control offences was influenced by the bridge. This later finding is not, at the present time, easy to rationalise and raises questions about the interpretation of the findings from the other regression models.

Discussion and findings of this paper raise questions that point to the need for new conceptual model at the regional level that go beyond traditional criminological theories. The regional scale implies, for instance, that the routine activity theory should be expanded to incorporate notions of mobility across borders (perhaps more complex than the ones applied to intra-urban patterns) for offenders, targets and potential guardians. The offenders’ perception of the international borders (and the risks involved in crossing it) is likely to be very different from other types of local barrier. All the characteristics that make an object a ‘suitable target’ (value, inertia, visibility and access) should be adapted to the context of borders. In this case, cross-border crime could be assessed as any other economic activity involving risk and reward. Rational choice theory should contribute to our understating of why certain offenders make a decision to cross borders. Notions of ‘capable guardians’ should take into consideration new forms of interaction in environments, which are well populated, but with a wide diversity of different forms of surveillance. This should assess the importance of formal and informal forms of guardianship (such as customs services, police patrols, CCTV cameras), that act in transport nodes (such as airports) and also in transitional places (such as trains, airplanes) to deter crime.

There is an urgent need for new understanding about crime at the regional level, particularly related to cases where geographical barriers are attenuated by investments in transport infrastructure. Within the European community, there have been in the 1990’s several initiatives devoted to reduce disparities among members’ states by strengthening the economic and social cohesion through investments in infrastructure. The Cohesion Fund created in 1993 and also the ESDP – European Spatial Development Perspective, 33 aimed at promoting a balanced access to infrastructure by having better accessibility (EC 1997, EC 1997). Future research should devote attention to understanding how the national and international dynamics of crime impact on crime trends at local and regional levels. This is particularly important when dealing with border regions, since they are more vulnerable to external influences than other regions. More needs to be done to investigate the relationship between cross-border/international organised offences and traditional crimes. An increase in smuggling of drugs generates not only other narcotic- related offences (e.g., selling) but also offences related to addiction (e.g., thefts, burglary).

The main challenge in assessing in more detail the effect (as here) of a new bridge is to combine individual or group level data on offence, victim and offender in both sides of the border. In this study, only aggregated data on offenders by citizenship for Swedish Öresund was available and was used to illustrate in offenders’ mobility. The individual data should refer to offenders that cross the bridge to commit crime as well as victims that are victimised in the other side of the bridge. Linking data on offence, victims and offenders is one of the core issues in modern criminology and findings of this study are an example of where the linkage is crucial to understanding process. However, still much work remains to be done in solving problems of data access and data comparability between countries. The fact that crime records in most countries are regarded as ‘very sensitive’ data at the co-ordinate level makes the process of acquiring the datasets long, difficult and costly. Combining crime data from different countries is an additional challenge. In several cases, offences are reported and recorded in different ways, which make comparisons between countries a difficult task (see Appendix 2).

There is also a need to control for the effect of infrastructure improvements at the border from other structural changes (particularly nationwide) that certainly affect crime levels and its geography. These factors act together impacting on crime and its geography. The increase in cases of human trafficking both in Denmark and Sweden is an example of how international trends have affected the region together with other factors that are more related to changes in the levels of policing and customs routines in the region. From national to local levels, police activity is regulated by structural conditions in each particular country, which is often directed to specific types of crimes. For instance, changes in taxes on products that are targeted by smugglers may affect the flow and direction of smuggling as well as the way such illegal trade is approached by the police. Thus, future research should be devoted to the effects of changes in political borders on crime. EU enlargement, which adds new member states to the existing fifteen, will certainly increase the need for this type of knowledge, particularly between the frontiers of the new states with the former Russian republics.

Acknowledgements This research was undertaken whilst Vania Ceccato was a visiting fellow at the Department of Geography at the University of Cambridge, England. The support of the Marie Curie Fellowship Scheme (Grant reference HPMF-CT-2001-01307) and STINT – The Swedish Foundation for International Cooperation in Research and Higher Education (Dnr PD2001-1045) are gratefully acknowledged by the authors. Thanks also to Dr Seraphim Alvanides, University of Newcastle for help in designing the 34 new spatial units in Z-DES. The authors would also like to express their thanks to the Copenhagen Police Authority in Denmark, Skåne Police Authority in Sweden, Malmö municipality and Skåne region for providing the dataset used in this analysis.

8. References Aebi, M, G Barclay, J Jehle, and M Killias. 1999. European sourcebook of crime and criminal justice statistics: key findings: Council of Europe. Ahlberg, J, and T Thedeen, eds. 1981. En undersökning avseende ekonomisk brottslighet inom tullens ansvarsomradet - ett pilotprojekt. Edited by D. Magnusson. Vol. 2. Stockholm: Utvecklingsenheten, BRÅ. Anderson, M, M den Boer, P Cullen, W Gilmore, C Raab, and N. Walker, eds. 1995. Policing the European Union. Oxford: Clarendon express. Andersson, A. E, and C. W Matthiessen. 1993. Öresundsregionen: Kreativitet, Integration. Copenhagen: Vaekst. Munksgaard. Anselin, L. 1992. SpaceStat Tutorial: a workbook for using SpaceStat in the analysis of spatial data. Morgantown: West Virginia University. Barclay, G, C Tavares, and A Siddique. 2001. International comparisons of criminal justice statistics 1999. London: Home Office. Bew, P, and G Gillespie. 1993. Northern Ireland: A Chronology of the Troubles 1968- 1993. Dublin: Gill & Macmillan. Bevelander, P, B Carlsson, and M Rojas. 1997. I kruksbärslandets storstäder: om invandrare i Stockholm, Göteborg och Malmö. Kristianstad: SNS. Block, R, and S Davies. 1996. The environs of rapid stations: a focus for street crime or just another risky place? In Preventing mass transit crime, edited by R. Clarke. Monsey, NY: Willow Tree Press. Brantingham, P J , and P L Brantingham. 1981. Notes on the geometry of crime. In Environmental Criminology, edited by P. J. Brantingham and P. L. Brantingham. London: Sage. Bygvrå, S, and H Westlund. 2001. Öresundsbro - ökad interaktion? en analys av inköpsresor m.m. före och efter brons öppnande. Aabenraa: Danish Institute of Border Region Studies. Bucken-Knapp, G, and M Schack, eds. 2001. Borders matter: Transboundary regions in contemporary Europe. Edited by J. Kuhl. Aabenraa: Danish Institute of Border Regions Studies. Cohen, L E, and M Felson. 1979. Social change and crime rate trends: a routine activity approach. American Sociological Review 44:588-608. Costello, A, and P Wiles. 2001. GIS and the journey to crime: an analysis of patterns in South Yorkshire. In Mapping and analysing crime data: lessons from research and practice, edited by A. Hirschfield and K. J. Bowers. London: Taylor and Francis. COWI. 2002. Trafikundersogelse på Oresund. Kongens Lyngby: DSB, Scandlines, Skånetrafiken, Öresundsbro Konsortiet. Di Nicola, A. 2001. Trafficking in migrants: a European perspective. In Cross-border crime in a changing Europe, edited by P. C. V. Duyne, V. Ruggiero, M. Scheisnost and W. Valskenburg. Huntington: Nova. Dunning, E. 2000. Towards a sociological understanding of football hooliganism as a world phenomenon. European journal on criminal policy and research 8 (2):141-162. 35

EC. European Regional Development Fund & Cohesion Fund 1997 [cited. ———. 2000. European Regional Development Fund and Cohesion Fund. European Commission 1997 [cited April 2000]. ———. 2000. European Regional Development Fund and Cohesion Fund. European Commission 1997 [cited May 2000]. EC, DG XVI. Spotlight on CohesionPolicy. Speeches of Commissioners 1998 [cited. Entorf, H, and H Spengler. 2002. Crime in Europe:causes and consequences. Berlin: Springer. Field, S , R V Clarke, and P M Harris. 1991. The Mexican vehicle market and auto theft in border areas of the United States. Security Journal 2 (4):205-211. Galeotti, M. 1995. Cross-Border Crime in the Former Soviet Union. Boundary and territory briefing 1 (5):1-26. Gehl, J. 1987. Live between buildings: using public space. New York: Van Nostrand Reinhold Company. Haining, R. 2003. Spatial data analysis: theory and practice. Cambridge: Cambridge University Press. Haining, R, and A Cliff. 2003. Using a Scan statistic to map the incidence of an infectious disease: measles in the USA 1962-1995. Paper read at Proceedings of the Geomed conference 2001, at Paris. Hajdnijak, M. 2002. Smuggling in Southeast Europe: The Yugoslav wars and the development of regional criminal networks in the Balkans. Sofia: CSD - Center for the Study of Democracy. Hjalmars, U, M Kulldorff, Y Wahlquist, and B Lannering. 1998. Increased incidence rates but no space-time clustering of childhood malignant brain tumors in Sweden. Cancer. HomeOffice. 2003. Routine Activity Theory. Home Office 2003 [cited May 2003]. Available from http://www.crimereduction.gov.uk/learningzone/RAT.htm Jamieson, R, South, N, I Taylor. 1998. Economic liberalisation and cross-border crime: the North American free trade area and Canada's border with the USA (Part 2). International Journal of the Sociology of Law 26: 285-319. Junninen, M , and K Aromaa. 2000. Professional crime across the Finish-Estonian border. Crime, Law and Social Change 34:319-347. Krajewski, K. 2001. Drug trafficking in Poland. In Cross-border crime in a changing Europe, edited by P. C. V. Duyne, V. Ruggiero, M. Scheisnost and W. Valskenburg. Huntington: Nova. Kuldorff. 1997. A spatial scan statistic. Communications in Statistics:Theory and methods 26:1481-1496. LaVigne, N. 1997. Visibility and vigilance: metro's situational approach ro preventing subway crime. Washington, DC: National Institute of Justice. Leimgruber, W. 1981. Political boundaries as a factor in regional integration, examples from Basle and Ticino. Regio Basiliensis 2 (192):192-201. ———. 1998. Border trade: the boundary as an incentive and an obstacle to shopping trips. Nordia 22:53-60. ———. 1999. Border effects and the cultural landscape: the changing impact of boundaries on regional development in Switzerland. In Nationalising and 36

Denationalising European Border Regions, 1800-2000, edited by H. K. a. J. Markusse. Amsterdan: Kluwer. Levine, N, and M Wachs. 1986. Bus crime in Los Angeles: I Measuring the incidence. Transportation reseach A 20 (4):273-284. Loukaitou-Sideris, A. 1999. Hot spots of bus stop crime. Journal of the American Planning Association 65 (4):395-411. Ma, J, R Haining, and S Wise. 1997. SAGE User's guide. Sheffield: University of Sheffield. Martin, D. 2002. Spatial patterns in residential burglary: assessing the effect of neighbourhood social capital. Journal of Contemporary Criminal Justice 18 (2):132- 146. Mattthiessen, C W. 2000. Bridging the Öresund: potential regional dynamics integration of Copenhagen (Denmark) and Malmö-Lund (Sweden), a cross border project on the European metropolitan level. Journal of transport Geography 8:171-180. ———. 2002. Öresundsomradet - Öresundsregionen: status 2001. In Öresundsforbindelse med ett hinder mindre: effekter pa integrationen vid avgiftsfrihet, edited by L. Soderstrom. Lund: Justnu. NCPD, and National Commissioner of Dannish Police. Status Report on Organised Crime in Denmark in 2001 2002 [cited. Available from (http://www.politi.dk/PUBL/Organiseret_kriminalitet_2000/Orgkrim_2001_endeligver sion_eng.doc. Openshaw, S, S Alvanides, and S Whalley. 1998. Some further experiments with designing output areas for the 2001 UK census. Paper read at ESRC/JISC Workshops planning for the 2001 census: determining academic community needs and strategy, at School of Geography. Osmonaliev, K. 2002. Legal and Organizational Aspects of the Fight against Money Laundering in Kyrgyztan. In Tackling Cross Border Crime: Challenges of International Development Co-operation. Bonn: The international Policy Dialogue. Osyka, I. 2001. Organised economic crime: problems in the Ukraine. In Cross-border crime in a changing Europe, edited by P. C. V. Duyne, V. Ruggiero, M. Scheisnost and W. Valskenburg. Huntigton: Nova. Persson, L G W. 1999. Cigarett smugglingens struktur, omfattning och utveckling under senare delen av 1990-talet: fran ax till limpa. Stockholm: BRÅ. ———. 1999. Cigarett smugglingens struktur, omfattning och utveckling under senare delen av 1990-talet: fran limpa till lunga. Stockholm: BRÅ. Poister, B. 1996. Transit-related crime in suburban areas. Journal of Urban Affairs 18 (1):63-76. Porter, M. 1996. Tackling cross border crime. London: Home Office. Rattcliffe, J H, and M J McCullagh. 1998. Aoristic crime analysis. International Journal of Geographical Information Science 12 (7):751-764. ———. 2001. Crime, repeat victimisation and GIS. In Mapping and analysing crime data: lessons from research and practice, edited by A. Hirschfield and K. J. Bowers. London: Taylor and Francis. Rikskriminalpolisen, RKP -. 1999. Öresundsforbindelsen: konsekvenser for svensk polis och tull. Stockholm: Rikskriminalpolisen - Kriminalunderrattelsetjansten. 37

Rosenfeld, R, S F Messner, and E P Baumer. 2001. Social capital and homicide. Social Forces 80 (1):283-309. Ross, C E, and S J Jang. 2000. Neighborhood dirsorder, fear, and mistrust: the buffering role of social ties with neighbours. American Journal of community Psychology 28 (4):401-420. Ruggiero, V. 2000. Transnational crime: official and alternative fears. International Journal of Sociology of Law 28:187-199. Ruyver, B, K Van Impre, and J Meese. 2001. The need for a multidisciplinary and proactive approach to human trafficking: reflections from two pilots. In Cross-border crime in a changing Europe, edited by P. C. V. Duyne, V. Ruggiero, M. Scheisnost and W. Valskenburg. Huntigton: Nova. SCB. Øresunds Databanken 2003 [cited. Available from http://www2.dst.dk/Öresund1/. Schhoenhardt, A. 1999. Organised crime and the business of migrant trafficking. Crime, Law and Social Change 32:203-233. Sherman, LW. 1990. Police crackdowns: initial and residual deterrence. In Crime and Justice: A Review of Research, edited by M. Tonry and N. Morris. Chicago: University of Chicago Press. Speer, D.L. 2000. Redefining borders: the challenges of cybercrime. Crime, Law and Social Change 34:259-273. SatScan:Software for the spatial and space-time scan statistics v.2.1. National Cancer Institute, Bethesda, MD. Stott, C, P Hutchison, and J Drury. 2001. Hooligans' abroad? Inter-group dynamics, social identity and participation in collective 'disorder' at the 1998 World Cup finals. British Journal of Social Psychology 40 (3):359-384. Tobias, J.J. 1967. Crime and industrial society in the 19th century. New York: Schocken. Tremblay, C R, and P Tremblay. 1998. Social structure, interaction opportunities and the direction of violent offenses. Journal of Research in Crime and Delinquency 35:295- 315. Tsebelis. 1989. The abuse of probability in political analysis: the Robinson Crusoe falacy. American Political Science Review 83:77-91. Ulrich, C. 1994. The growth of crime in Russia and the Baltic region: RFE/RL Research Report. Vagg, J. 1992. The borders of crime: the Hong Kong-China cross border criminal activity. British Journal of Criminology 32 (3):310-328. Van Houtum, H. 2000. An overview of European geographical research on borders and border regions. Journal of Borderlands studies 15:57-83. White, R C. 1932. The relations of felonies to environmental factors in Indianapolis. Social Forces 10:498-509. Wikström, P O. 1991. Urban crime, criminals, and victims: the Swedish experience in an Anglo-American Comparative Perspective. Stockholm: Springer-Verlag. Wiles, P, and A Costello. 1999. The "road to nowhere": the evidence for travelling criminals. Sheffield: Police Research series. Wilson, R. 2003. No Frontiers: North-South integration in Ireland 2003 [cited 16th June 2003]. Available from (http://cain.ulst.ac.uk/dd/report11/report11k.htm). 38

9. Appendices

Appendix 1 - The border between Sweden and Denmark: the Öresund region

The Swedish-Danish border is a sea border, divided by the sound of Öresund. Despite the fact that Öresund divides Sweden and Denmark, it has created over the centuries a contact area that has fostered relations and conflicts between Danes and Swedes on either side. Öresund has not always been the border between Sweden and Denmark. During the Viking times, wherever ‘the blurred borders’ were, Danes and Swedes shared practically the same language and fought for parts of the same territory. Until 1657, the entire Öresund region was part of the Kingdom of Denmark. In the war of 1657-1658, the Swedes defeated the Danish and the southern part of Sweden, Scania, became part of the Swedish Kingdom. Since then, the region developed separately for nearly four hundred years and it was not until twenty-five years ago that mutual administrative, political and economical effort was combined to bring Öresund together as an integrated economic region.

Table 1 Area, population and municipalities of Öresund region Denmark Sweden Total area, Km2 9.883 11.194 Total population (1999)* 2.394.765 1.123.786 No. Municipalities 102 33 average area, Km2 97 339

Source: SCB, Orestad: statistik om Öresund, 2003. *For comparison, Denmark’s total population is 5,1 million people and Sweden’s total population is 8,8 million.

Öresund includes Copenhagen metropolitan area (Hovedstadsregionen), the capital of Denmark, and Scania (Skane) the Southern part of Sweden. The latter includes the third largest city in Sweden, Malmö (Table 1). Thus, in this study, we exclude the low-density populated parts of the Danish Öresund, such as V Sjælland, Lolland and Bornholm

Before the bridge, Öresund was linked by boats. In 1999, there were 160 departures in each direction every day for ferryboats carrying goods and passengers and 50 daily departures for hydrofoil passenger boats. Despite the fact ferryboats still connect part of the region, the boat route services between Malmö-Lund to Danish Öresund was taken out of after the bridge’s opening. The price of taking a trip either by train or by car over the bridge is relatively expensive taking into consideration the same distance trip in other parts of the region (Soderstrom, 2002). A one-way train ticket over the Öresund bridge costs 70 DKK (around £6 or $10) whilst the same ticket for a similar distance between Lund and Landskrona in Sweden costs 37 DKK (around £3 or $5). For cars, there are two methods of payment. The automatic method involves the payment is performed in advance and an electronic transmitter is attached to the car window that sends a signal to the control station when the vehicle passes the border. The second method involves the traditional payment using money or credit cards at the time when the border is crossed.

Despite the relativly small differences in language (Sjoberg, 2001) and in the political and administrative systems, Sweden and Denmark differ in terms of regulations and legislations. There are structural differences in, for instance, income taxe, tax on products, social security and health care between the two countries. This difference in taxation particularly on certain products between Denmark and Sweden (e.g., alcohol) acts as a stimulus to border crimes such as smuggling.

Öresund, as other parts of Scandinavia, has a strategic place in wider regional organised crime. According to Galeotti (1995) Scandinavia has little experience in dealing with the wave of criminality coming from Russia and certain Baltic countries. The Swedish, Finnish, German, Estonian, Latvian and Lithuanian bank notes have all begun circulating within the Baltic/Nordic economies, along with refugees, drugs, stolen 39 guns and Russian oil. The Baltic States have become important way stations for the illegal export of metal and minerals, largely from Russia. Poland has a central role, since many illicit goods flow from and to Eurasia, across Poland, for example, case for stolen cars, drugs smuggling, prostitution and racketeering.

Criminal legislation in the Nordic countries is similar and for many decades has been specially protected by internal agreements. This means in practice that if an offender from Sweden commits a minor crime in Denmark (e.g., burglary), he/she will be charged in Denmark go to court in Sweden and has the right to take his/her sentence in a Swedish prison. For drug related crimes, however, the offender goes to court in Denmark but still may have the right to choose where he/she wants to serve the sentence. In Öresund, there is a co-operation between the Police, Custom Services and coast police patrol. 40

Appendix 2 – Methodological considerations

There is a debate whether or not it is feasible to use official crime statistics for international comparative scientific research or policy driven analyses. This because nations differ widely in the way they organise their Police and court systems, the way they define their legal concepts and the way they collect and present their statistics. The lack of uniform definitions of offences, of common measuring instruments and of common methodology makes comparisons between countries a difficult task (Aebi et al. 1999). One solution is to make comparisons of trends over time of recorded crimes as suggested by (Barclay, Tavares, and Siddique 2001), instead of comparing absolute numbers. Since the Danish and the Swedish code system differ (the total number of crime codes is 350 in Sweden whilst in Denmark it is 365), we adopt a standard definition and the two Police Authorities were invited to indicate which were the offences that best fitted this pre-established standard. One typical example of this difference is the case of residential burglary. In Sweden, residential burglary is recorded as ‘burglary followed by theft’ (codes 9801 and 9802) whilst in the Danish code criminal system, burglary in residences (codes 75212 and 75213) is recorded separately from ‘theft in residences’ (codes 75584 and 75585). One way to partially solve this problem is to combine the statistics for these types of crime in Denmark and later make the comparison. The comparison is then focused on trends over time-space and not on absolute numbers.

Another challenge when working with crime data from different sources is creating a compatible database, in other words, having data at the same spatial level. Crime data over Sweden does not have a georeferrenced location in space as in the case of the dataset for Denmark. The dataset required therefore geocoding over 300.000 addresses for each year. Geocoding is the process of matching records in two databases: the crime address database (without map position information) and a reference street map or other "address dictionary" (with known map position information). Thus, the quality of the geocoding process depends very much on the quality of the crime records, the quality of the address dictionary, which the geocoding is based on (x,y co-ordinates) and the chosen method for geocoding.

In terms of quality of the crime records, the first problem found was the variable level of detail either on place (where event took place), or the time that it happened or both. Some crimes are easier to record accurately than others. The typical example is residential burglary, is reported in detail (street and street number). On the other extreme are acts of vandalism in public places that are not reported in great detail (e.g., parks, parking lots, transport links, commercial areas). Often crime records are incomplete. People when visiting an unknown neighbourhood may not know where they were when they were robbed. A similar reasoning explains the problems found when recording the precise time of the offence. Car related thefts is a good example since reporting might be done several hours after than the event took place. A “range in hours” (e.g., 12:00-14:00) is sometimes used when nobody knows the exact time that the event occurred and aoristic modes of data transformation (Rattcliffe and McCullagh 1998, 2001). There are also cases when the crime site is unidentifiable, for example when it took place between A and B, on a train, bus, train, airplane, or through the Internet. In the case of Skåne database, it is unclear in some cases if the address, when reported, is actually the one where the individual was victimised. This may create ‘false hotspots’, since a high number of records may be reported in airports, bus stations and railway stations.

Geocoding also depends on the quality of the reference database containing each pair of co-ordinates, which in its turn is dependent on how up-to-date the addresses are, and how compatible with the structure of the existing crime records each address is. In the Swedish case, the mismatch between crime record and background database has restricted the automatic geocoding in GIS to only 30 per cent of all crime records, demanding an exhaustive process of data -cleaning. The quality and quantity of these addresses is also affected by the size of the urban area. In small towns there are places that do not have a limited boundary or even a street or number attached to a known geographical area, which makes both the crime record and the geocoding process very unreliable. The method chosen used x,y-coordinates and zone boundaries in a GIS desktop Mapping System. In order to improve the matching rate in the geocoding, the addresses of known ‘hot spots’ (train, bus stations, schools) were gathered using Internet searching tools or collected from each municipality. 41 42

Appendix 3 - Characteristics of the dataset Type of data Description Year Source Offences Type Swedish code Danish code Smuggling1 4002, 4022, 4003 76501, 71282 1998- Skåne Policy Pick pocketing 845 75607, 75609, 75611, 75613, 75551 2001 Authority Vandalism 1201-3, 1205,1207 77111-5, 77121 Molestation 403, 407, 408, 410 74265, 74266, 74271 Copenhagen Vehicle theft2 801-806 75411, 75421, 75441, 77271-77, 77279, 77311 Policy Residential burglary3 825-826,843-844, 857, 874 (9801-9802 in 2001) 75211-75213, 75311 Authority Narcotics 5004-5006, 5010-5012 71281, 84110-84111, 75312-75313 Assault 351-358, 361-368, 371-378,381-388 73215-16, 73241, 73251, 73261-73262 Swedish Personal robbery 877-879, 890, 894-895 (2001) 76319, 76329 Customers Shops related offences 864, 870, 818, 853 76313, 76323, 75152-75153, 75560,75562-3 Services Theft from cars 840 75512-75514 Crime against common order 1602-1607 70421, 70423 Robbery in banks and post offices 862-863, 868-869 763114 Offences in places of tourism 886-887 75114, 75184 Domestic violence 381-384 - Fraud 903-906 - Demographic and Proportions of socio-economic Population younger than age 18 (P18, DP18) 2000 Malmö indicators Population older than age 65 (P25, DP25) 2000 municipal Unemployed labour force (PUnemp, DPUnemp) 2001 database Population with (at least) a parent born abroad (Ppfo, DPpfo) 2001 (Statistics Population born abroad (Pfore, DPfore) 2001 Sweden) Population moving into the area (Pin, DPin) 2000 Population moving out the area (Pout, DPout) 2000 Region Rental public housing apartments (PPho, DPho) 1990 Skåne Apartments owned by private companies (PPc, DPPc) 1990 Privately owned single family houses (Pvill, DPvill) 1990 Families average disposable income (PInc, DPInc) 1999 Households receiving social allowance (PSal, DPSal) 1999 Local leisure associations by pop (PLass,DPLass) 2002 ‘Neighbourhood Watch Schemes’ by pop (PNws, DPNws) 1998 Land use Dummy for urban and urbanised areas (Urb, DUrb) Skåne traffic indicators Dummy for university Campus and surroundings, 1st order neighbours (Univ, DUniv) authority Dummy for commercial areas and surroundings, 1st order neighbours (Shop, DShop) Dummy for central/railway station and surroundings, 1st order neighbours (Stat, DStat) Swedish Dummy for hospital and surroundings, 1st order neighbours (Hosp, DHosp) National Dummy for Centre (Centre, DCentre) Road Dummy for areas of tourism attraction, (Turis, DTuris) Associa-tion Proportion of buss stops by population (PPuss, DPbuss) Access to the bridge - Inverse distance to bridge (Acess, DAcess) 1. For Sweden, the data is incomplete, data on seized goods from the Customers Services was used instead; 2. This excludes bicycle, boat, other; 3.This includes cellar, attic, leisure house for Sweden; 4.This does not include post offices; 5.For Sweden, this data includes theft or burglary in churches, museums, sport arenas, cinemas, hotels whilst for Denmark, the category chosen was “Burglary in general, others” (Indbrud i øvrig, inkl. kirke/museum).