The emergence of a “city of cages” in Lima: neighbourhood appropriation in the context of rising insecurities

LSE has developed LSE Research Online so that users may access research output of the School. Copyright © and Moral Rights for the papers on this site are retained by the individual authors and/or other copyright owners. Users may download and/or print one copy of any article(s) in LSE Research Online to facilitate their private study or for non-commercial research. You may not engage in further distribution of the material or use it for any profit-making activities or any commercial gain. You may freely distribute the URL (http://eprints.lse.ac.uk) of the LSE Research Online website.


Introduction
Lima, like other Latin American metropolises, has experienced profound changes in the context of broader transformation processes since the 1990s (e.g. Díaz, 1997;Pradilla, 1998;de Mattos, 2002;Portes and Roberts, 2004). Greater Lima (Lima Metropolitana) is now a mega-city with approxi-mately 8.1 million inhabitants (INEI, 2005: online). The functional structure of the Peruvian capital has undergone a process of decentralisation expressed in the emergence of additional nodes of specialised formal and informal activities (Chion, 2002;Ludeña, 2002). On the other hand, it seems that the segregation patterns identified for Latin American cities until the 1980s have remained largely intact (Bähr and Mertins, 1995). Some evidence suggests however that they are gradually replaced by a more complex structure on the micro-level (Joseph, 2004: 17), a process that has often been described as spatial fragmentation (e.g. Pradilla, 1998).
Gated residential developments have emerged as one characteristic feature of the new spatial order of Latin American metropolises. Yet, they vary considerably throughout the region regarding dominant types, significance, urban historic origins, and regulation by the state (e.g. Caldeira, 2000;Svampa, 2001;Cabrales, 2002;Hidalgo et al., 2003;Meyer-Kriesten et al. 2004). This article analyses the process of residential fortification in Lima Metropolitana, where a local newspaper has observed the emergence of a "city of cages" (La República, 22.06.2003). Lima represents a unique case with some distinctive features. Subsequently enclosed neighbourhoods are by far the most important type of residential enclave. 3 They are characterised by a high degree of informality, the dominant role of the residents and their spread across the metropolitan area and crosscut socioeconomic lines, among other aspects.
The process of enclave building involving the appropriation, control and fortification of residential areas in Lima will be analysed in the wider context of a profound transformation since the 1990s. It will be argued that the practices of spatial intervention can be linked to the persistence of social inequalities, the more recent confrontation with a range of insecurities and the inefficiency of the state.

"The new rules of the game"
Since the worldwide economic recession during the 1970s, the Latin American economies were drawn into a series of crisis. The situation escalated in the "lost decade" of the 1980s. In Peru, the economic crisis was exacerbated by the political crisis due to the interior conflict. To overcome this situation and to reactivate their economies, most Latin American governments opted for structural adjustment policies under the surveillance of the international financial institutions such as the International Monetary Fund and the World Bank. In Peru a phase of authoritarian internal policy and orthodox structural adjustment was inaugurated under the Fujimori government elected in 1990.
The new economic development model was one of the most comprehensive implementations of the recommendations made by the "Washington Consensus" (Parodi, 2001: 302;Thorp, 1999: 228).
This shift had significant social impacts. Despite the overall economic growth since the 1990s, the high inequalities dividing the Peruvian society have been manifested (Figueroa, 1996;Gonzáles, 1998;Riofrío, 2004). It is estimated that only ten per cent of the households were able to improve their economic situation (Ugarteche, 1998: 157). The traditional middle-class was particularly hardhit (Gonzáles, 1998: 86). The share of households below the poverty line rose slightly to 54 per cent in 2000 (Joseph, 2004: 37).
The social impact of the "reform packages" can be illustrated by the consequences for the labour market. Many formal sector jobs were lost. Trade liberalisation caused the collapse of several formerly protected industrial sectors that could not compete under free-market conditions (Joseph, 1999: 46). With the objective of stabilising the budget, the public sector was downsized and many of its companies privatised (Gonzáles, 1998: 117). Furthermore, the deregulation of the labour market caused an increasing instability of employment conditions. In Lima, the percentage of the work-force in the informal sector increased from 48 to 58 per cent between 1989(Equipo Ciudad, 2003. The "change in the rules of the game" (Grompone, 1999: 81) has confronted the majority of the population with a wide range of insecurities regarding the ability to maintain their standard of living and social status. The withdrawal of the state further deepens the exclusion of large parts of the population from adequate housing and basic infrastructure, health services and education (Ríofrio, 2004). The reproduction of social inequalities is aggravated by a decomposition of the social (Lechner, 1998). It can be argued that these overall insecurities find their most explicit expression in the increasing fear of becoming a victim of crime.

Responses to urban insecurity
This chapter focuses on the interaction between the insecurity caused by crime and the response by the security providing actors. It is difficult to measure the impact of crime in Peru. The official data is not representative of the actual situation. It is estimated that only ten per cent of all crimes are reported to the police (INEI, 1998b). The lack of reliable information reinforces the influence of inter-personal and media discourses on shaping public opinion (Plöger, 2006b: 73). The most comprehensive evidence of the impact of crime on society has been collected by two, albeit slightly outdated, studies about victimization in Lima (INEI, 1998b;Apoyo, 1999). According to these surveys the majority of Limenõs perceives urban life as increasingly insecure. More than a third of the respondents indicated having become victim of some form of crime in 1997 (ibid.).

Security gap and security actors
From the perspective of the population, the state is viewed as unable to deliver a sufficient level of protection. Those institutions responsible for law enforcement, police and judicial system, are the least trusted of all public institutions (Basombrío, 2003: 62). Various authors have elaborated on the issue of institutional weakness in Peru and argued that it needs to be interpreted as a historical consequence of incomplete nation building and the inherent social inequalities (e.g. Figueroa, 1996;Matos, 2004;Crabtree, 2006). The proliferation of informal practices -including those used to appropriate residential areas -can be interpreted in this context. In the realm of security, it can be argued, that the insufficient service provision by the state has caused a "protection gap". As a result, additional public, private and community-based actors have emerged in most districts. In many areas this has resulted in a complex structure of security providing actors.
In Lima, the Peruvian National Police (PNP) employs a force of approximately 23,000 in 166 police stations (IDL, 2004: 62). They are distributed unequally. The protection of an area depends upon its location and socio-economic status. In general, the ratio of police officer per inhabitant decreases towards the peripheral and, mostly, poorer areas. While the wealthiest district, San Isidro, has a ratio of 1:190 inhabitants, it exceeds 1:1,000 in the peripheral districts (Plöger, 2006b: 75). 4 Three quarters of the 49 districts of Lima Metropolitana have now responded to the growing demand for citizen security by establishing their own security force. These so-called serenazgos are basically "quick-response units" aiming at closing gaps in the security provision. They do not possess the same competences as the police, though. In 2004, this force employed approximately 5,000 in Lima (El Comercio, archive). Similarly to the police, their number and equipment varies between districts. Due to higher tax revenues, wealthier districts dispose of more resources to provide such additional services. Furthermore, private security services have experienced a rapid growth since the 1980s. Although official information is not available, it can be estimated that more than 40,000 security guards are employed in Lima (Basombrío, 2003;El Comercio, archive). The majority has not received any job training and is employed on an informal basis. Their increasing number has been interpreted as a sign for the privatisation and informalisation of security provision (Joseph, 2004: 54). Private security guards are usually employed by a group of residents on neighbourhood or street-level. Again, their distribution correlates with socio-economic status, as their number is higher in middle and upper income areas.
Some neighbourhoods have formed community-based security groups (e.g. comités de autodefensa, rondas vecinales, juntas vecinales). For the purpose of this brief overview they will be referred to as neighbourhood watch groups. In contrast to the actors mentioned above, they are more common in poorer and peripheral neighbourhoods that are under-policed and where the lack of financial resources does not permit the employment of private security guards. Reliable information about their number was not available.
A dramatic sign for the failure of the state to provide with security in some areas is the increasing number of incidents of "popular justice". In the first ten months of 2004 alone, almost 700 attempts of lynching were officially reported in Lima Metropolitana, mostly in marginal areas (La República, 04.11.04). According to a survey conducted at the same time, 64 per cent of the interviewees agreed with the notion that "the population has the right to take justice into its own hand" if the public authorities are unable to provide sufficient protection from crime (El Comercio, 15.11.2004).

Reaction of public authorities
This section discusses how the state has responded to the increased demand for seguridad ciudadana (citizen security) and to the proliferation of practices that undermine constitutional rights such as the free access to public spaces. Three efforts need to be mentioned in particular.
The first is the reform of the police and the judicial system under the Toledo government between 2001 and 2006. These reforms were not directly related to the demand for citizen security but rather aimed at restructuring those public institutions that were deeply involved with the corrupt Fujimori regime. Though some progress was achieved, the reforms of the two crucial institutions for law enforcement have been regarded as incomplete at best (Rospigliosi, 2006).
The second policy approach was the ratification of a framework for improved citizen security in 2003. Citizen security was defined as "the integrated action developed by the state, with the collaboration of the civil society, aimed at guaranteeing the peaceful coexistence, the obliteration of violence and the pacific use of public streets and spaces, while at the same time contributing to the prevention of criminal offences" (Congreso de la República, 2003). 5 This law aimed at establishing formal links between public institutions and residents. According to this scheme, the residents form juntas vecinales to cooperate with the police and local authorities in order to receive improved policing. The junta vecinal identifies for example crime hot spots and communicates them to the authorities. Though there is little evidence about the impact of these partnerships so far, surveys suggest that the population is relatively satisfied with this approach of community policing (Basombrío, 2003).
The third response is the attempt to regulate the -mostly unauthorised -closures that the residents have installed on public streets. Throughout the 1990s the remit was with the metropolitan authority of Lima. Due to insufficient resources it was not able to control the compliance with existing regu-lations, though. It was during this phase that the "security landscape" in Lima became increasingly informalised. Most district authorities are not interested in confronting their residents (and voters) with unpopular measures such as the removal of gates, especially when they do not dispose of the appropriate means to guarantee sufficient levels of security in their jurisdictions. Nevertheless, several local authorities -predominantly of districts with a high number of informal street closures (e.g. La Molina, San Borja, Los Olivos, Ate) -published their own regulation. In 2004, the Peruvian office of the ombudsman presented a study that not only documented the irregularity of the situation and the negative impact on public spaces but also acknowledged the demand for improved security and the inefficiency of the responsible institutions (Defensoría del Pueblo, 2004). This resulted in a new legal framework by the metropolitan authority later that year aiming at formalising and regulating the security devices installed on public streets (MML, 2004). The district authorities have implemented this ordinance very slowly however. Only a few consolidated districts have made efforts to enforce the legal guidelines, while the informal status quo persists in the remaining districts (El Comercio, 26.07.2005).

Control and appropriation of residential areas
The article argues that local spatial practices, such as the formation of guarded and gated residential areas in Lima, must be understood in the context of rising overall insecurities. Harvey (1996: 293) has interpreted such reactions on the local level as an attempt to guarantee the permanences of places against the penetration of external and spatially unbound forces of change. A continuously transforming urban environment hence requires constant adjustment. In areas that experience problems with crime and social disorder, the inhabitants tend to search for solutions to reduce the level of threat. In order to re-establish a feeling of "spatial comfort" and personal safety, the residents try to gain stronger control over their surroundings (Madanipour, 2003: 52).
One approach of illustrating the process of enclave building has been the concept of "condominisation" (Plöger, 2006b: 22). It describes how residents appropriate their neighbourhoods using different forms of spatial intervention. As a result of these practices, the neighbourhood becomes superimposed by a constellation of layers of control. Three such layers can be identified: The physical layer includes physical (e.g. gates, barriers and fences) and technological (e.g. alarm systems) measures as well as the employment of security personnel. The realisation of these interventions requires the availability of financial resources. It is the most obvious layer of control, both for residents and outsiders. The organisational layer includes both the organisation of residents (e.g. through residents associations) as well as their collective attempt to establish certain rules regarding the use of neighbourhood spaces (e.g. through internal regulations). The cultural-symbolical layer works on a more subliminal level. Its effect on the accessibility of a space is indirect. Spatial appropriation is achieved for example through the use of "spatial signalling devices" (Suttles, 1972: 161), which can include signs such as message boards, posters or graffiti. It might also be expressed by spatial behaviour, for instance the use of spaces by the local community. As will be seen below, the impact of the individual layers on the accessibility of an area varies considerably.

Residential enclaves in Lima Metropolitana
Closed residential complexes in Lima can be traced back until the 19 th century (Ludeña, 2004). Quintas, for example, are ensembles of residential units built around a private cul-de-sac, whose only entrance is often equipped with a gate. Still, a significant expansion of security-related measures in the residential sector occurred late in comparison with other Latin American metropolises.
During the mid-1980s the terrorist movement Sendero Luminoso increasingly focussed its activities on Lima. Confronted with this threat, the middle and upper classes that had lived relatively undisturbed until then started to move into guarded apartment complexes (condominios de edificios) and gated residential developments (barrios cerrados). The massive spread of gated residential enclaves occurred only since the 1990s however and was closely related to the fear of crime.
Two general types of residential enclaves can be distinguished in Lima Metropolitana: subsequently enclosed neighbourhoods and private residential developments ( fig. 1). Their main difference is the origin of the security-related measures. Subsequently enclosed neighbourhoods are existing neighbourhoods that are transformed into residential enclaves through the continuous exertion of spatial control and the implementation of security-measures by the residents. Therefore they can also be called "ex post condominiums". Private residential developments or "real condominiums" are already planned and developed including security measures and the division between private and communal spaces. This type subdivides into horizontal developments such as "typical" USstyle gated communities (Blakely and Snyder, 1997) and vertical developments such as apartment complexes.

Measuring the spread of residential enclaves
This article presents evidence from a study about the fortification process in Lima, which focussed on subsequently enclosed neighbourhoods (Plöger, 2006b). As mentioned above, the emergence of fortified residential areas has been studied broadly for other Latin American metropolises. For Lima no such research had been conveyed yet.
Official data on residential enclaves was not available. Apart from a few wealthier districts, local authorities had almost no information about street closures in their jurisdictions. The only sources of secondary information were a selection of articles from the archive of El Comercio, the most influential Peruvian newspaper. Due to the lack of information, the author created an inventory of residential enclaves in Lima Metropolitana in 2004. This was achieved through on-site visits in all districts, the available information from newspapers and conversations with local experts. This enabled an estimate about the distribution and number of residential enclaves. In accordance with another source (El Comercio, 26.07.2005), there are approximately 3,000 street barriers and 300 residential enclaves in Lima Metropolitana (Plöger, 2006b).

Map 1: Residential enclaves in Lima Metropolitana
Content: J. Plöger; cartography: P. Sinuraya With the exception of denser and more central areas, residential enclaves have spread over most parts of the metropolitan area (map 1). The dominance of subsequently enclosed neighbourhoods is evident. Eight districts with a high proportion of residential enclaves can be identified: La Molina, Los Olivos, Santiago de Surco, San Miguel, Cercado de Lima, Chorrillos, San Borja and San Luis. The relatively wealthy district of La Molina has by far the highest number of street closures. In 2004, an inventory by the district authority has counted 529 gates, barriers and other devices (information provided by the district of La Molina). In general concentrations occur in the proximity of residential areas of lower socio-economic status, traffic thoroughfares or areas with nonresidential land use (e.g. industrial or commercial). In comparison with other Latin American metropolises, "typical" gated communities are relatively seldom and limited to a few wealthy, suburban areas mostly in La Molina and Santiago de Surco. "Vertical" condominiums are located either as exclusive condominios de edificios in older and denser upper-class districts such as Miraflores and San Isidro or as more simple developments for the (lower) middle-classes in the adjacent traditional middle-class areas. The fortification process is crossing socio-economic lines. This is a particularly noteworthy phenomenon because residential enclaves are usually attributed to middle-and upper-income groups. Even in marginal areas in peripheral districts such as Comas, San Juan de Lurigancho or Villa El Salvador residents have installed physical devices and organised some form of vigilante control (Plöger, 2006a). The distribution of gated neighbourhoods however correlates with income and is therefore more likely to occur in wealthier areas.

Subsequently enclosed neighbourhoods: a profile
One main objective of the study was to understand residential enclave building in Lima in its wider social context. Special attention was given to the practices of spatial control used by the residents. For this purpose 21 subsequently enclosed neighbourhoods were chosen as case study areas. The inventory enabled the geographic representativeness of the selection. Additionally, a set of indicators was used to guarantee also a representative socio-economic distribution: the average socioeconomic status according to micro-level data from the National Statistics Institute (INEI 1998a); 6 lot sizes; average property value for the lot including the typical housing type for the area; and the observed level of consolidation and equipment of the neighbourhood. In each neighbourhood, indepth interviews were carried out with the representative of the local organisation that was responsible for the implementation of security-related interventions, mainly the installation of street closures and the employment of security guards. The following section presents the main characteristics of the selected case study areas. 7 Their defining characteristic is that they have been subsequently equipped with the different layers of control mentioned above. This has been achieved through a collective effort by the residents or an organised group of residents such as the original homeowner or residents associations, juntas vecinales, or specialised committees that had evolved around a specific task like security provision.
The fortification process is a relatively recent phenomenon in Lima. In the majority of the case study areas, the gates or barriers were only installed since 2000, in half of the cases only since 2003. In these cases the measures were described as a direct response to increased insecurity, mainly caused by crime. In many cases the loss of spatial control was also attributed to specific social groups. These were often labelled as "gente de mal vivir", which translates into "people with bad lifestyles" and includes, for example, alleged gang members, drug consumers, beggars or petty criminals. In only four cases the street closures were installed at an earlier date. In two peripheral neighbourhoods they were a response to insecurity deriving from the terrorist threat in the late 1980s; the other two areas were affluent neighbourhoods where the street closures were installed in the 1970s and 1980s to guarantee an exclusive lifestyle.
Subsequently enclosed neighbourhoods are further characterised by the high degree of informality of the measures. In less than a quarter of the case studies the gates and barriers were authorised by the district council. The neighbourhood representatives mentioned a variety of reasons for not having such an authorisation: In three cases the procedure to obtain the permission was described as too bureaucratic, slow, expensive, or complicated; in four areas it was claimed that a formal permission would not be required due to a good relationship with the local authority; in five areas the representatives remained inactive because legal action against them was highly improbable; in the remaining four areas the situation had not been discussed yet.
The quality and quantity of the measures depends on the available financial resources and therefore the socio-economic status of the neighbourhood. The employment of security guards is a good example to illustrate this connection. In marginal settlements, scarce financial resources often do not permit the employment of security personnel, but in two such neighbourhoods the security guards were recruited from among the residents. In poorer neighbourhoods the monthly contribution for security services does usually not exceed the equivalent of 2 Euros per household. In the heterogeneous group of middle-income neighbourhoods the monthly contribution per household varies between 2 and 10 Euros per household, while upper-income neighbourhoods were able to charge between 10 and 30 Euros.
Another important factor is the degree of community organisation. In combination with the financial resources it determines the accessibility of an area and the degree of spatial control by the residents. In the majority of the case studies the residents had achieved perimeter access control through the installation of street closures and/or the employment of security guards. In the remaining six areas, the security measures were organised only on street-level. Here, community cohesion and neighbourhood organisation were generally weaker. This can be illustrated, again, by using the example of the security guards. To guarantee a security service in the long-run requires some degree of community organisation. Only in a few affluent areas over fifty percent of the households paid their monthly fees for the security service on a regular basis. Due to low contribution rates some areas were even forced to abandon the employment of guards. Depending on his employment status and skills, a guard charges between 90 and 200 Euros monthly. The installation of a gate amounts to 450 to 900 Euros only once. For this reason, some areas with a lower degree of internal organisation managed to install gates but were not able to sustain the employment of guards.
In the following two chapters, the fortification process will be illustrated by presenting evidence from one lower and one middle class neighbourhood.

Enclave building in lower-class areas
In Lima, the fortification process has extended to poorer and marginal districts. This is also true for Villa El Salvador, which has previously received attention due to its development as a "planned marginal settlement" since the 1970s (Bähr and Klückmann, 1985). Recently, several neighbourhoods in the district have been equipped with gates and barriers. The highest number of street closures has been installed in Barrio II in the large Pachacámac area. Barrio II is a vast neighbourhood with more than 13,000 inhabitants that mostly belong to the upper lower class (INEI, 1998a). The neighbourhood was founded to provide lower-level employees of the public sector with affordable housing. Most of the area was equipped with basic infrastructure financed through a government programme and international development aid. The simple housing modules constructed since 1986 allowed for future extension. A quarter of the area was illegally invaded, though, and developed as a marginal settlement with self-constructed houses. Despite the initial support, the area is not fully consolidated today. It has a failing basic infrastructure, lacks street paving and public lighting and many lots have an uncertain property situation.
Due to its many problems of marginality, there has always been a functioning neighbourhood association in Barrio II. The thematic focus of this organisation has however shifted over time. According to the president of the neighbourhood association, security has become the most important task in the last years. Crime was mentioned as a serious problem in the area. Especially robberies and muggings would be frequent.
The insecurity was attributed not only to the inhabitants of adjacent neighbourhoods, but also to the inhabitants of some "dangerous" areas inside Barrio II (see map 2). These areas match with those that had been invaded illegally. In the words of the neighbourhood representative, the informal practice of land invasion was mirrored in a general disrespect for social norms, disorderly social behaviour and delinquency. Furthermore, the informal market areas and the vacant land reserved for parks were perceived as dangerous. The situation was exacerbated by the fact that both National Police and municipal serenazgo were not properly attending the area: "Pachacámac is a critical area. The police response to crime is ineffective and random, due to their lack of personnel. They [the police] are not able to guarantee satisfying security, although this area has serious problems with criminals and other «people with bad lifestyles»." (president of neighbourhood association, October 2004) This statement reflects the experience of inadequate security provision of many neighbourhoods in the predominantly poor periphery of Lima Metropolitana. With one police officer per approximately 1,500 inhabitants, Villa El Salvador is one of the most under-policed districts (El Comercio, archive). The police own only 12 vehicles to patrol the jurisdiction of 350,000 inhabitants. In reaction to the increased public demand for security, the local authority has inaugurated a serenazgo in 2003. But, the units are patrolling mainly in the commercial zones of the district. As a consequence of the threat of crime and the inefficiency of the public security provision, the residents have started installing street closures in different areas throughout the neighbourhood since 2001. So far, a total of 31 gates and two barriers have been erected on block-level (see fig. 2). These obstacles serve mainly to restrict the access of vehicles that would be necessary to transport stolen goods. Pedestrians can usually pass through a side opening in the gate. Due to the uncoordinated installation of measures, the "security geography" of Barrio II resembles a labyrinth (see map 2). The residents of the respective streets were able to install the devices by charging a fee from the households. In some cases the funds could be raised through the organisation of local festivities. In none of the cases had the residents applied for a municipal authorisation for the security measures. At the time of the investigation, security guards were not employed. The neighbourhood representative said that earlier experiences had shown that the service would be too expensive in long-term. Furthermore, there had been allegations of some of the guards being involved in burglaries. The residents were discussing the option of revitalising a neighbourhood watch scheme used during the years of terrorist activities.
Although, the number of robberies has apparently decreased in the gated street blocks, the representative emphasised that many other areas in the neighbourhood, especially the market areas and open spaces as well as those areas originating from land invasions remained dangerous.

Enclave building in middle-class areas
Although lower-income areas are increasingly being fortified, it is especially the middle-income areas that have become subsequently enclosed. One of the main concentration areas of middle-class residential enclaves is situated in the Western extreme of the district of Ate, which is very heterogeneous in terms of socio-economic structure and land-uses. In this part, almost all neighbourhood subdivisions have been subsequently equipped with security measures.
The case study area belongs to the larger neighbourhood of Recaudadores. The different subdivisions of Recaudadores have been built by large public-sector companies since the last half of the 1960s for some of their employees. The individual subdivisions comprise a few blocks. These are usually arranged around a central green space (parque) that provides the semi-official name for the respective residential area. The selected case study is referred to as "residential area Parque Piura". This neighbourhood consists of 80 lots and has approximately 440 residents that belong predominantly to the "typical" middle class (INEI, 1998a).
The original neighbourhood association of the area had disintegrated in the 1990s due to a lack of participation and the absence of significant problems affecting the whole community. In 2004, the only functioning residents' organisation was the junta vecinal. Security provision was therefore the only task based on a common interest. The junta vecinal was formed by a group of residents to improve the security situation in the area. In order to receive better policing they made efforts to establish closer links with the local police station and the district authorities. Still, their main focus was the implementation of security measures in the neighbourhood.
The sensation of insecurity was not directly linked to crime, but originated from the presence of certain social groups that were associated with a range of immoral, harassing, and sometimes criminal behaviours. These groups were generally referred to as "gente de mal vivir" or "people with bad lifestyles". Their presence in the open spaces (parques) in and around the case study area was perceived as a threat to the integrity of the neighbourhood. The representative of the Parque Piura area argued that it was the free access to the public spaces that would attract these unwanted groups: "The parque was in a condition of total neglect. It was badly illuminated at night. It was frequented by gente de mal vivir, for example drug addicts and couples of dubious reputation that did not belong to the neighbourhood and had no reason to be here [...]. But, we did not have the authority to say to these people: «Please leave this place!» Why? Well, unfortunately the parques are public spaces and all persons therefore have the right to use them." (representative of junta vecinal, November 2004) From the perspective of the residents, this feeling of exposure to insecurity and loss of spatial control was aggravated by the lack of security provision from the police and the serenazgo. The police for instance would not do enough to re-establish order in the public spaces. As a response to this situation, the junta vecinal decided to install gates at the six street and two pedestrian entrances to the neighbourhood in 2001 (see map 3). Cars can access the area through the only gate that remains open during daytime. Pedestrians can still use side entrances. To guarantee that the gates equally affect all areas, a different gate is used every two months (see fig. 3). The neighbours have also installed an acoustic alarm system consisting of two sirens that can be used to warn the other residents if necessary. The costs for the street gates amounted to 750 Euros each and 200 Euros for each of the two gates at the pedestrian entrances. To cover these expenses, every household was asked to pay a special fee of approximately 60 Euros. The physical security measures were legalised by the district authority of Ate in 2004. It must be noted that this district had published its own regulation that allows street closures if they are based on a neighbourhood security plan. The junta vecinal has also organised the employment of three security guards. Two guards, working in 12-hour shifts, are permanently controlling the central green area. The third guard is employed at night to control the accessible gate. Each household is asked to contribute 10 Euros monthly to pay for the security service. According to the representative of the junta vecinal, about 90 per cent of the households are regularly paying that fee. Compared with most of the other case study areas, this is a very high participation rate.
The representative of the junta vecinal stated that the security situation had improved considerably. Crime had been reduced to almost zero. This was mainly attributed to the fact that the gates and guards would now prevent unwanted groups from entering the neighbourhood.

Conclusion: Lima a "City of Cages"?
In the context of broader political, economic and social restructuring, Latin American metropolises have been transformed considerably since the 1990s. Gated residential areas can be interpreted as a characteristic feature of the emerging "neo-liberal city" (Pradilla, 1998: 199).
This article argues that the formation of residential enclaves in Lima occurs as a consequence of the intertwined relationship between the exposure of the population to rising insecurities and the inefficiency of the state and its institutions to provide satisfactory services. For many households, a wide range of insecurities increasingly threatens the livelihood, employment conditions and incomes, social status and access to services. One pronounced outcome of the instability of an unequal society is the confrontation with direct insecurities deriving from the real or perceived threat of crime and deviant social behaviour. This was clearly evident in both of the presented case studies.
In many areas, the population has responded to the "security gap" by establishing residential enclaves through the appropriation, control and fortification of their neighbourhoods. This process is driven by the residents, hence the dominance of the particular type of the subsequently enclosed neighbourhood. The applied practices of spatial intervention can be interpreted as an attempt to reinstall a feeling of spatial comfort. In comparison with other services, security can be exerted relatively easy on neighbourhood level. The "security geography" of an area is the physical outcome of these interventions. It depends upon three factors: the available financial resources and therefore the socio-economic status of the area; the importance of the local security discourse; and the degree of community cohesion and residential organisation.
The appropriation of neighbourhoods occurs as part of a profound informalisation of Peruvian society. Matos (2004) has argued that the incomplete consolidation of the state and its inability to alleviate social problems has resulted in a desborde popular ("popular overflow") expressed by the use of strategies outside of the state's power sphere. The formation of an increasingly informal "security landscape" is further enhanced by the quasi-legitimation of the status quo by public authorities who are unable to enforce existing regulation.
The formation of residential enclaves indicates a further problem, namely the spatial reproduction of social inequalities in Lima. Socio-economic status determines the ability of an area to react to the "security gap". As has been shown, wealthier areas receive a more comprehensive police service.
The widespread perception of insecurity has caused the proliferation of additional security providing actors. The security forces employed on district level reinforce the pattern of socio-economic inequality. More affluent districts dispose of higher funds to pay for the extra layer of protection. The distribution of private security guards also correlates with income. Due to the insufficient policing of their areas, the residents of poorer neighbourhoods are often self-organising security provision, for example through the creation of neighbourhood watches.
Different attempts have been made by the authorities to improve the security situation. These are however not seriously challenging the socio-spatial segmentation of the urban structure resulting from the informal appropriation of neighbourhoods. Any sustainable attempt to create an urban environment with a higher quality of life for all sectors of the population needs to address the origins of residential enclave building, especially the persistence of socio-economic inequalities and the inefficiency of public institutions.