Arrival Infrastructures: Segregation of Displaced Migrants and Processes of Urban Change in Athens

Abstract

Recent treatment of accommodation and arrival infrastructures for asylum seekers and refugees has fuelled international research on refugee reception policies in urban environments and on the consequences of related initiatives of the European Union and international organizations such as the UNHCR. Using Athens as a case study, this article provides empirical evidence to revive the theoretical treatment of the importance of arrival and accommodation infrastructures in urban areas in transition. We collected and compiled data from four sources: the 2011 population census, the 2018 ESTIA accommodation program and the UNHCR Site Management Support (SMS) Reports of Temporary Accommodation Sites and Reception and Identification Centres (RICs), and a primary survey of services for asylum seekers and refugees. After the geocoding of data, we calculate indices for key dimensions of the segregation of accommodated asylum seekers and foreign nationals. We discuss the findings, seeking to highlight how the location and the composition of accommodation infrastructures has been influenced by a wider process of urban change and adaptations to global forces, leading not only to the transformation of inner-city zones but also suburban and peri-urban areas.

1. Introduction

The significance of arrival neighbourhoods and zones in transition has been established since the early founding texts of the human ecology of the Chicago school. In Burgess’s (1928) seminal articles “The Growth of the City” [1] and “Residential Segregation in American cities” [2], transition connotes two interacting processes vital for segregation: a spatial process, i.e., change in interstitial areas to accommodate the competition for residential, business, and industrial use of land, and a social process, i.e., transition in the life course and the assimilation of migrants through intra-urban mobility. The transformation of “areas in transition” was considered to be part of a concentric pattern of urban expansion, and assimilation was anticipated to initiate when migrant newcomers arrived at the “ports of first entry” looking for employment and affordable housing and mobilising their support networks to confront prejudice and resistance of the established local communities to their settlement.

Subsequent criticism of both the concentric development and assimilation model has led to the elaboration of complex urban social ecology patterns and has revealed that entry points are not only found in inner-city areas [3,4]. Moreover, the process of assimilation is not unidirectional from inner-city to suburban areas nor is it shaped by sedentary preferences and permanent residency. Contemporary mobility and migration trajectories are complex despite states’ efforts to control mobility by prioritising security and cultural homogeneity.

These drawbacks of human ecology have recently been addressed by the “infrastructural turn”, according to which urban infrastructures are socio-technical systems produced by material, social, and symbolic practices [5,6]. International studies on reception and integration policies viewed urban accommodation schemes as part of wider urban infrastructures and discussed the implications of the location in central, suburban, or peri-urban areas [7,8,9,10]. Indeed, a recent research strand on arrival infrastructures draws upon critical urban theory and the work of Lefebvre to elaborate how infrastructures of solidarity sustain the co-habitation and claims of the displaced to the right to the city, centrality, and participation in urban life as opposed to their marginalization in urban peripheries [11,12].

The revival of the theoretical discussion on arrival infrastructures should be set in a historical context when forced migration and protracted displacement on a global scale has attracted policy and research attention since the 1990s. UNHCR’s Policy on Alternatives to Camps [13] has been acknowledged to be a paradigm shift, challenging the inadequate camp-based policies, and—combined with the UN’s ‘Adapting to an Urban World’ project—has highlighted the need of humanitarian organizations and states to ‘adapt’ their reception and refugee integration policies to urban spaces. The distinctive UN influence on European policies has been evident since the escalation of the so-called refugee crisis in 2015, especially in countries of the European South like Italy [14] and Greece [15].

Although the influence of UN to EU initiatives has been a vital source for planning the improvement in arrival infrastructures and the livelihoods of asylum seekers and refugees, the benefits of urban accommodation schemes have in many cases not been materialised. Significant policy studies have shed light on the controversial effects of dispersal policies and decentralised asylum systems in Greek, Italian, and German cities that received the bulk of migrants [16]. On the one hand, the inadequate supply or the poor-quality accommodation units in peripheral or deprived city areas, the hostility of local governments and local anti-migrant mobilizations, and the privatization and disassociation of support services from mainstream provisions combined with the deterrence of settlement and mobility restrictions to produce conditions of protracted displacement (e.g., for Italy, Ambrosini (2021), Annunziata (2020), Campesi (2018), Lumley-Sapanski (2022), Semprebon and Pelecani (2019) [17,18,19,20,21]; e.g., for Germany, Kreichauf (2023), Bernt et al. (2022), El-Kayed et al. (2020) [22,23,24]; examples from Greece are referenced in subsequent sections). On the other hand, the relative autonomy of migrant-friendly local authorities and their co-operation with international agencies and civil society organizations has had some success in integrating accommodation to support services, mobilising a variety of interethnic networks and offering opportunities for inclusion (e.g., for Italy, Ambrosini (2021), Boano and Astolfo (2020) [17,25]; e.g., for Germany, Kreichauf and Mayer (2021), Marcińczak and Bernt (2021) [26,27].

Nevertheless, the majority of such policy studies are disassociated from the long-established research on ethnic segregation and do not use advanced methods of a spatial analysis with notable exceptions (e.g., Gerten et.al. (2023) [28] in Dortmund, Marcińczak and Bernt (2021) [27] in Berlin, Salah et al. (2019) [29] in Turkey, Taubenböck, Kraff, and Wurm (2018) [30] with a comparative methodology on the morphology of arrival neighbourhoods) possibly because small-scale data were not available, especially in countries with deficits in the administration of migration statistics.

This article aims to address part of this gap by using Athens as a case study and by compiling a variety of data sources as explained in Section 2, Materials and Methods. Being inspired by critical urban theory and the work of Lefebvre, we suggest to theorise transition as a process of differentiation, which not only concerns city concentrations but also extended transformations in areas far beyond the dense population centres, and to investigate how arrival infrastructures may contribute to the uneven thickening, and stretching, of the urban fabric [31,32,33]. Yet, we also suggest that the diversity of outcomes of arrival infrastructures on segregation can be better understood when considering how contextual factors shape how the flows of capital, i.e., globalization from above, and the flows of people, i.e., globalization from below, will be accommodated in space. In this sense, the impact of arrival infrastructures on the segregation of migrants is context-dependent because their location and functions are determined by the combined effect of institutions and collective struggles involving the market, the state, civil society, and durable urban forms and materiality [34].

There are four good reasons to use Athens as a case study of the role of arrival infrastructures in the urban geography of forced migration under conditions of a global economic crisis and hesitant recovery in the metropolises of the European South [35].

First, the “long summer of migration” in 2015 marked a change in the role of south-eastern European cities located on the new migration routes from the Middle East. The closure of the Balkan route, and the EU–Turkey joint statement in 2016, signify a turning point in the Greek state’s reception policies and contradictory efforts to adapt to international humanitarian protection rules [36]. In this context, Athens emerged as a major European gateway of forced migration.

The reception system expanded rapidly to include two types of accommodation infrastructure alongside the implementation of the hot-spot approach in the Aegean islands. On the one hand, Open Temporary Reception Facilities (“Sites”) were created in isolated areas in mainland Greece, suburban and peri-urban areas in Athens, which were turned into sites of prolonged displacement [37]. On the other hand, urban accommodation in apartments, hotels, and other buildings in the city of Athens, and other Greek cities, was a social innovation introduced through a specially designed program (ESTIA). ESTIA (Emergency Support to Integration and Accommodation) (2017–2022) was a program of housing vulnerable asylum seekers in the urban fabric under the auspices of UNHCR, EU funding, and implementation by municipalities and NGOs, and has aspired hopes for establishing a distinctive form of social housing by reserving a pool of private properties for affordable renting [15,37,38]. Two competing rationales shape the establishment and management of ‘sites’ (camp-like) and urban accommodation infrastructures as a response to global changes, the management of the Great recession, and the inflow of displaced migrants [36]. ‘sites’ are official spaces of containment and are a means for exercising state territoriality and for establishing an internal border. Urban accommodation involves a remodelling of market relations and regulations, with the engagement of the local administration and the civil society, so as to meet humanitarian aims and the prospects of social integration. The ESTIA project was significant in advancing the humanitarian rationale; it was discontinued in 2023 but has influenced the shaping of a similar housing project (HELIOS-Hellenic Integration Support for Beneficiaries of International Protection) for those asylum seekers who received a refugee status.

Second, Athens presents an ideal example to study forms of micro-segregation in arrival areas. The growth of the Athenian metropolis and middle-class suburbanization continued apace through the post-war period and until the sovereign debt crisis, resulting in high levels of social mix in central neighbourhoods. The acute production of housing since the late 1950s and up to the late 1980s, initially seeking to give a solution to the extreme needs of internal migration, led to an intensive and expansive growth of the urban space that succeeded to accommodate the inflow of migrants from eastern Europe, from the early 1990s to the mid-2000s. Despite the relatively low level of the segregation of migrants from Greeks, an ethnic hierarchy emerged, especially across central, inner, and outer suburban areas often and around the low-quality housing stock left empty by the Greeks [39,40,41]. In parallel to the middle-class suburbanization and littoralization of Athens, vertical–social and ethnic segregation became a dominant feature of the city’s social structure. This form of micro-segregation is to be found in the Athenian apartment buildings, the ‘polykatoikies’. Increased homeownership combined with lack of social housing resulted in the cohabitation of upper middle classes with low-income households, and Greeks with foreigners in one single building [42,43,44]. Recently, numerous projects placed vertical segregation in the spotlight, questioning micro-segregation in different social and ethnic contexts [45,46,47,48,49].

Third, Athens is well suited to study the combined effects of the economic and the so-called migration crisis in arrival areas of the European South, since the Greek sovereign debt crisis has not only exacerbated the existing spatial inequalities but has also deepened deprivation on central and peripheral areas as a result of metropolitan shrinkage combining with the decline in migration from the Balkans and natural population growth in core city areas, and divergent land use changes in peri-urban areas [50,51,52].

Last but not least, the city has historically exhibited moderate or low levels of social class segregation, conforming to the so-called inverse Burgess model, due to the distinctive features of Southern Housing and Welfare Regimes, family provisions, and self-built housing [53,54]. However, in line with comparative research on segregation in southern Europe, we question the capacity of historically prevalent family provisions and homeownership to compensate for the lack of public housing and consider whether the availability of an informal and residual rental stock to asylum seekers in central and peripheral areas shaped their temporary settlement and segregation [55].

Within this context, we use measures of segregation and mapping to provide an empirical answer to the following two questions:

• Have camp-like and urban accommodation facilities contributed to change intensity and key dimensions of ethnic segregation in the Athens metropolitan area? Specifically, do they exhibit a pattern of dispersal or concentration that changes the existing spatial concentration of migrant communities and their disadvantaged positioning in housing markets?
• Are camp-like and accommodation facilities adequately linked to services across the metropolitan space as to create an accessible network of social infrastructures?

The research design is described in Section 2, Materials and Methods. The results and related maps are presented in Section 3, Results. In the presentation of results and in Section 4, Discussion, we draw from insights and commentary given during research workshops with officials from local government, accommodation agencies, and employees of support services to offer some insights on how contextual factors have contributed in shaping reception areas and their transformation. The conclusions highlight the international significance of the results, the limitations of the present study, and its potential to inspire similar comparative studies.

2. Materials and Methods

2.1. Research Design

To answer the two questions set out above, the research design relied on collecting, combining, and geocoding data from four main sources. Then, we applied a spatial analysis, first to detect the possible changes that accommodation infrastructures had on dimensions of segregation related to the centralization, concentration, and dispersal of migrants, and second to detect if urban accommodation and service provisions concentrated within areas of vertical segregation. The explanation of the results of the spatial analysis was enriched by feedback during workshops with local policy makers and accommodation agencies. The research design is schematically presented in Figure 1 and explained in subsequent Section 2.2 and Section 2.3.

2.2. Data Sources and Geocoding

Data were collected from four main sources:

• The 2011 population census to detect patterns and dimensions of segregation already established before the “long summer of migration”;
• The January 2018 ESTIA project data to assess the links between the spatial attributes of the refugee settlements and the established patterns of segregation;
• The January 2018 UNHCR Site Management Support Reports of Temporary Accommodation Sites and Reception and Identification Centres [56] (henceforth, SMS reports) seeking to investigate the influence of such accommodation facilities, planted by the administration, on the limits of the urban fringe. In 2018, the UNHCR published 7 such reports officially stating on their cover that they seek to “allow for better planning and to address gaps where highlighted”.
• A primary survey of providers of support services to migrants undertaken in December 2020.

The data analysis was complemented by 3 workshops with researchers, local government bodies, non-governmental organizations, and employees of support services to explore which factors shaped the design of the ESTIA program and its impact on the formulation of reception and social integration policies.

The 2011 census data were analysed and mapped on the lowest possible spatial level, a modified version of 2011 Census Tracts, the Urban Analysis Units (henceforth, URANUs). This dataset, concerning the ethnic composition of Athens, was provided by the Panorama of Greek Census Data 1991–2011 [57] and cuts down the metropolitan area in 3.000 URANUs of an average population of 1250. The use of this detailed version of the city, applied to all the datasets of this work, ensured trustworthiness of the compilation of spatial analysis indicators and better understanding of the urban patterns.

The ESTIA project data concerning the nationality of the beneficiaries, the type of accommodation unit, and the capacity and the occupancy per unit were provided by the UNHCR. After being geo-localised, the information was linked to the URANUs’ base-map in order to be mapped. This procedure ensured confidentiality; the 1802 initial dwellings were aggregated in 713 URANUs, helping comparisons with the rest of the datasets.

The third source used for this work is the January 2018 UNHCR SMS report comprising information about the composition of the four sites—Schisto, Eleonas, Skaramangas, and Lavrio—established in the broader metropolitan area of Athens. The data are organised per site and linked to the URANUs’ level, providing information about the nationality of the accommodated populations.

All three datasets were organised either on the exhaustive categories of nationality or on more general ones like the geographic regions or even the development level [58]. Concerning the ethnic composition of Athens, individuals from the eastern European countries dominate the urban space (

Table 1. Ranking of the most important non-Greek nationalities in Athens. Census and Accommodation projects.

National Census (2011)				ESTIA and Sites (2018)				ESTIA		Sites
Rank	Country	Pop.	%	Rank	Country	Pop.	%	Pop.	%	Pop.	%
1	Albania	216,504	49.7	1	Syria	7059	44.0	5014	45.0	2045	41.6
2	Pakistan	23,340	5.4	2	Afg/stan	4240	26.4	2690	24.2	1550	31.6
3	Romania	21,221	4.9	3	Iraq	2681	16.7	1712	15.4	969	19.7
4	Bulgaria	19,597	4.5	4	Iran	560	3.5	321	2.9	239	4.9
5	Georgia	12,767	2.9	5	Palestine	310	1.9	237	2.1	73	1.5
15	Syria	5885	1.4
17	Afg/stan	4673	1.0
23	Iraq	2983	0.7
30	Iran	1147	0.3
41	Palestine	538	0.1
	Other	141,998	10.4		Other	1200	7.4	1163	10.4	37	0.8
	Total	435,636			Total	16,050	100	11,137		4913

Source: EKKE-ELSTAT, 2015; UNHCR-ESTIA project, 2018; UNHCR SMS report, 2018; data processed by the authors.

) with Albanians representing almost 50% of the non-Greek nationals, being followed by Romanians (4.9%), Bulgarians (4.5%), and Georgians (2.9%). The Pakistani community is the sole exception to this pattern, representing almost 6.0% of the foreign nationals. The ESTIA project and the sites’ ethnic composition, as expected, show great similarities since they incorporate individuals from the Middle East (Syria and Iraq) or the Greater Middle East (Afghanistan). By 2011, the nationalities linked to the migration wave had established small communities in Athens, not standing out in the overall ethnic composition of the city. They are ranked between the 15th and the 30th place with Syrians being the larger group representing 1.4% of the total foreign population. In absolute numbers, if we consider both ESTIA and sites, these communities change profoundly and their presence increases by at least 50% (Syria + 120%, Afghanistan + 91%, Iraq + 90%, Palestine + 58%, and Iran + 50%). Moreover, the inflow of displaced migrants together with the outflow of migrants from the Balkans and population shrinkage [59] might have had stronger effects than the ones this paper estimates.

Implementing more than 1800 residences, accommodating more than 11,000 migrants (4211 families), and having the capacity to accommodate more than 12,500 people unveil the scale and complexity of the ESTIA project. The scale explains how the main factor that could produce spatial differentiations in their placement was in fact the supply of properties. The social and ethnic composition at the micro-scale is dependent on the capacity of local submarkets to adjust to new demand. The polikatoikia that preserved the spatial proximity, independently of social and ethnic statuses, in Athens seems to provide the solution once again. It is indicative that certain apartment buildings accommodated more than one family, revealing a new form of ethnic mix. To explore spatial associations between the localization of accommodation units and the pre-existing Ethnic Vertical Segregation pattern of the city, we proceeded to a detailed analysis and mapping.

The last source used in this analysis was produced between September and December 2020 and registers the services’ network linked to the needs of migrants. Through extensive online research, we located 546 services provided by 47 organizations. Interviews followed to verify and document in detail information about the service providers, the location, the type of services, and the profile of users. That way, we managed to verify and map the provision of 353 services on offer during 2020 by 23 organizations.

2.3. Segregation Analysis: Indices and Maps

Ethnic segregation patterns were identified in two steps. The first step consisted of the calculation [60,61] of three spatial analysis indicators, the Index of Dissimilarity (ID), the Absolute Centralization index (ACE), and the Relative Centralization index (RCE). The calculation formulae of the software we used [60] are available online at https://doi.org/10.4000/cybergeo.12063. The centralization index shows to which degree a group is located near the centre of an urban area. The index reflects the extent to which a group is spatially distributed by distance from the CBD compared to the distribution of land around the CBD. The index varies between +1 and −1, with positive values indicating a tendency for a particular migrant group to reside close to the city centre and negative values indicating a tendency to live in outlying areas. A score of 0 means that the migrant group has a uniform distribution throughout the metropolitan area [62]. The Relative Centralization index, as has been computed here, measures the extent of a migrant group’s centralization relative to the total population. It varies from −1, when the members of a migrant group are located further from the city centre than the total population, to 1, for the opposite situation. When the index is equal to 0, the migrant group and the total population have the same spatial distribution around the city centre. The dissimilarity index varies between 0 and 1.0 and conceptually it represents the proportion of minority members that would have to change their area of residence to achieve an even distribution, with the number that would have to move under conditions of maximum segregation [62].

The analysis aims to detect how the uneven distribution—the dissimilarity and concentration of the most prevalent groups of displaced migrants—change when taking into consideration the population residing on the sites, therefore on the outskirts of the city, and the beneficiaries of the ESTIA project, mostly residing in the central neighbourhoods of the city. First, we calculated dissimilarity and concentration indices for 2011 census data and then we repeated the procedure adding the SMS and the ESTIA datasets. We have not calculated measures of exposure (isolation and interaction) because existing literature on Athens has verified that migrant groups with a non-EU origin are not isolated mainly due to their small share in the total population [63].

Given that most residences, almost 99%, that accommodated migrants through the ESTIA project are in apartment blocks, the second step of the analysis explores the links between vertical segregation, a form of micro-segregation found in high-rise areas of apartment blocks in Athens, and the distribution of the ESTIA residences.

As traditional mapping techniques fell short in capturing verticality, it remained unexplored until recently. Introduced by L. Leontidou in 1990 [54] and evidenced by T. Maloutas and N. Karadimitriou (2001) [42], it was finally measured and mapped by T. Maloutas and S.N. Spyrellis in 2016 [43], after the introduction of floor-level information at the census tract level in the 2011 census survey for the first time.

The centre of Athens is composed of a relatively young and homogenised housing stock. In these parts of the city, the predominant residential strategy is homeownership in apartments although it also constitutes the main area for the rented housing market. These apartment blocks built on small plots hold an average of 15 apartments in five to seven floors. According to the 2011 census, 90.2% of the population resided in apartment blocks while only 6.4% lived in buildings constructed before 1961—the stock was mostly built between 1960 and 1980 (65.7%) through the antiparochi system. This system of a private initiative was based on a land-for-flat triangulated relation between an owner of a plot, a small-scale constructor, and the future buyers. This system was bucked by the state in order to compensate for the absence of social housing. Research has shown that an important part of the high-rise urban stock in Athens is characterised by vertical segregation, following a segregation mechanism, on the micro-scale, common in southern European cities between others [45].

Vertical segregation is related to important inequalities on housing quality across floors on this, otherwise homogenised, housing stock. The lower the apartment, the higher the disadvantage due to poor light, housing amenities (heating or glazing), noise, size of apartments, and absence of a view. The occupational status, homeownership, and ethnicity are important elements of differentiation since tenants seem to reside in smaller apartments in lower floors while the upper, sunnier, and bigger apartments are kept for their owners. Migrants, from less developed countries, are to be found in lower and less desired residences and under-represented on top floors [43].

The ESTIA dataset does not disclose information concerning the floor of the accommodating apartments; therefore, it was not possible to directly assess the extent to which the beneficiaries were vertically segregated. To overcome this issue, we mapped the Ethnic Vertical Segregation (henceforth, EVS) using the 2011 census data and then compared the results to the spatial distribution of ESTIA dwellings. To our knowledge, this is the second attempt to analyse and map EVS in Athens, the first made by Maloutas and Spyrellis in 2016 [43].

For the analysis and mapping on the EVS in Athens, we followed a method introduced by Maloutas et al. in 2022 [44]. First, we located the URANUs dominated by a high-rise susceptible to vertical segregation building stock, i.e. URANUs where the residents in apartment blocks account for at least 67% of the total population (average of residents in apartment blocks in the metropolis) and where those living in the upper and the lower floors account for more than 30% (the second criterion is used in order to avoid small apartment buildings of three floors often found in the suburbs). We identified as such 1134 units out of 3000 (37.7%).

Then, we proceeded to the identification of vertically segregated areas on one hand (simultaneous over-representation of Greeks and citizens from developed-economy countries on upper floors and over-representation of foreign citizens from developing-economy countries, in comparison with their average in the metropolitan area, in the lower floors) and areas of vertical advantage and vertical disadvantage on the other hand. The concentrated vertical advantage (SAI) (Standardised Advantage Index (SAI) = (% of Greeks and citizens from developed-economy countries (4th or higher) in URANU i/% of Greeks and citizens from developed-economy countries on disadvantaged floors (basement or ground floor) in URANU i)/(% of Greeks and citizens from developed-economy countries on advantaged floors (4th or higher) in the study area/% of Greeks and citizens from developed-economy countries on disadvantaged floors (basement or ground floor) in the study area)) is corroborated when the share of Greeks and citizens from developed-economy countries divided by their share on lower floors in an area is considerably greater than the respective fraction in the metropolis. Concentrated disadvantage (SDI) (Standardised Disadvantage index (SDI) = (% of working-class categories on disadvantaged floors (basement or ground floor) in URANU i/% of citizens from developing-economy countries on advantaged floors (4th or higher) in URANU i)/(% of citizens from developing-economy countries on disadvantaged floors (basement or ground floor) in the study area/% of citizens from developing-economy countries on advantaged floors (4th or higher) in the study area)), respectively, is confirmed when the share of foreign citizens from developing-economy countries on lower floors divided by their share on upper floors is considerably greater than the respective fraction in the metropolis. In this way, we distinguish areas where the two ethnic categories are very unequally distributed among advantaged (upper) and disadvantaged (lower) floors. By combining SAI and SDI, we identified the EVS profiles expressed through 5 groups of URANUs mapped further down.

3. Results

3.1. Segregation of Displaced Migrants: Intensifying the Centre–Periphery Dynamics

Table 2. Urban segregation indices. Census and Accommodation projects.

Data	Albania	Middle East	Greater Middle East	Northern Africa	Indian Peninsula
ID 1 Census	0.2859	0.5113	0.5045	0.6625	0.5933
ID Census and ESTIA		0.5256	0.5075	0.6606	0.5909
ID Census and Sites		0.5468	0.5201	0.6626	0.5925
ID Census and ESTIA and Sites		0.5500	0.5211	0.6607	0.5901
ACE 2 Census	0.6395	0.7675	0.8126	0.7516	0.5564
ACE Census and Estia		0.8055	0.8228	0.7538	0.5581
ACE Census and Sites		0.7221	0.8040	0.7516	0.5572
ACE Census and Estia and Sites		0.7684	0.8144	0.7538	0.5589
RCE 3 Census and Sites	0.1407	0.3217	0.4601	0.4077	0.1275
RCE Census and Estia and Sites		0.4090	0.4765	0.4102	0.1285
RCE Census and Sites		0.2128	0.4357	0.4078	0.1289
RCE Census and Estia and Sites		0.3203	0.4536	0.4102	0.1300

¹ Index of Dissimilarity, ² Absolute Centralization index, ³ Relative Centralization index. Source: EKKE-ELSTAT, 2015; data processed by the authors.

shows the values of the dissimilarity, the Absolute Centralization, and the Relative Centralization indices, which have been calculated for different groups of ethnicities, adding each time to the census 2011 data the newcomers accommodated in facilities of the ESTIA project and sites. To provide a comparative picture, the indices have been calculated for Albanians, who form the largest share of the migrant population that arrived and settled in Athens during the 1990s, and do not constitute a group of ESTIA beneficiaries or sites.

The findings in Table 2 suggest that the urban accommodation infrastructures for asylum seekers and refugees did not contribute to increasing the segregation of specific ethnic groups but only slightly in the case of newcomers from the Middle East (cf. Syrians). The forced migrants who recently arrived at the metropolis have different points of entry and their presence in the city is more segregated than the Albanian population who settled and dispersed throughout the whole metropolis during earlier periods. The dissimilarity index increases from 0.5113 to 0.5256, i.e., only by 2.8%, when the ESTIA beneficiaries from the Middle East are included in the estimations. Changes in the dissimilarity index are close to zero and negative for all other ethnicities, suggesting that their placement included areas different from the residential locations of their co-ethnics during the census. However, the placement of displaced newcomers strengthened the central concentration of ethnicities from the Middle East; the ACE index rose by 5% (from 0.7675 to 0.8055) and the RCE index rose by 27.1% (from 0.3217 to 0.4090) when the ESTIA beneficiaries were included in the estimation. A similar, but weaker change, applies to ethnicities from the Greater Middle East (cf. Afghans) for which the ACE index rose by 1.3% and the RCE index rose by 3.6% when the ESTIA beneficiaries were included in the estimation. Changes in the centralization of migrants from northern Africa and the Indian Peninsula are negligible. This change becomes more evident in the northern neighbourhoods of the city of Athens, Piraeus, and the adjacent areas (see Figure 3).

A different effect of the location of sites in Eleonas, Schisto, Skaramangas, and Lavrio on segregation is evident, as has been expected. The dissimilarity index increases by an extra 4.8% (i.e., from 0.5256 to 0.5500) when persons from the Middle East contained in sites are included in the estimations. The increase in the values of the dissimilarity index for ethnicities from the Middle East due to their containment in sites is more than twice the increase reported for ESTIA beneficiaries. The dissimilarity index also increases by an extra 2.7% (i.e., from 0.5075 tο 0.5211) when persons from the Greater Middle East contained in sites are included in the estimations. The increase in the values of the dissimilarity index for ethnicities from the Greater Middle East due to their containment in sites is almost five times the increase reported for ESTIA beneficiaries. Changes in the dissimilarity index are close to zero and negative for all other ethnicities. The values of the centralization indices reflect the peripheral location of the majority of sites (see Figure 1). Table 2 shows a significant decrease in the values of the Absolute and the Relative Centralization index concerning, firstly, ethnicities from the Middle East (cf. Syrians) who were transferred from the country borderlands to the most distant locations and, secondly, ethnicities from the Greater Middle East (cf. Afghanis).

Overall, urban accommodation schemes seem to have contributed to retaining segregation levels at moderate and manageable levels for the recently arrived displaced populations in Athens. Urban accommodation has contributed to strengthening their concentration in central city areas with already plural ethnic composition, fittingly called by some human rights activists ‘arrival neighbourhoods’. Sites and camp-like provisions contribute to significantly increasing segregation levels especially due to the containment of the Syrians and Afghanis in peripheral locations on the city edges and beyond.

3.2. Infrastructures of Forced Arrival and Peripheral Segregation

From the extensive literature on refugee camps as means for containment and segregation, Kreichauf’s comparative research (2018) is most relevant to the Athenian context. Kreichauf (2018) [64] coined the term ‘infrastructures of forced arrival’ to highlight the stabilization of temporary, enlarged, remotely located, and spatially isolated camps with lowered living standards. Qualitative research in Athens has revealed the detrimental effects of the location and the physical structures of camps on access to basic services and on eroding the liveability and autonomy of asylum seekers [37,64] and is compatible to our quantitative findings.

Our analysis relies on UNCHR reports about four sites that were operating in the Athens metropolitan areas in 2018 (Figure 1). These include one in Eleonas, at a small distance from the city centre but in a non-residential/ex-industrial zone, one towards the south in Lavrio, a former mining town, and two more in the industrial areas of Schisto, a former military base close to an informal working-class settlement, and Skaramangas, a pier of a port-town with shipyards, on the western outskirts of the city. Four more sites are located in the remote administrative borders of the Attika region but have not been included in our analysis. Had we included them in the analysis, segregation and decentralization indices would have increased. The recent history of urban camps in the metropolitan area of Athens goes back to 2016 when thirteen sites operated as an emergency response to decongest the Aegean islands, to offer temporary protection, and to cope with the basic needs of those trapped in the country after the closure of the Balkan corridor. Whilst the initial location of the thirteen sites included areas close to transportation infrastructures (ports, train stations), the remaining sites in operation today are located in abandoned military bases, factories, or warehouses at a considerable distance from the city centres in degraded and environmentally hazardous areas [65]. Decisions for their location did not involve a consultation process with local authorities but only ex-post negotiations to circumvent mobilizations against their establishment [65].

The analysis of UNCHR reports (Figure 2) suggests that in the urban fringe and the peri-urban space, the ethnic composition of migration sites was largely determined by the state or the international organizations through massive referrals from border spaces and the Greek islands. To a lesser extent, and especially in the first two years of their operation, NGOs and migrant networks played a role in directing newcomers to the sites. In January 2018, all sites were under the responsibility of the Greek Authorities. For the Athenian sites, the Hellenic Navy, the Hellenic Army, the Ministry of Migration Policy, The Reception and Identification Service (RIS), or the International Non-Governmental Organizations (INGOs) were responsible for their management, either individually by one authority or by joined forces.

In the following table (

Table 3. Ethnic composition of sites (2018).

Camp	Syria	Afg/stan	Iraq	Iran	Pakistan	Other	Total
Lavrio	1.23	0.46	0.55	1.75	0.00	2.36	1.00
Schisto	0.34	2.80	0.30	1.73	0.00	0.00	1.00
Skaramagas	1.27	0.34	1.98	0.57	0.00	0.00	1.00
Eleonas	0.66	1.93	0.36	2.58	3.56	0.00	1.00
Sites’ total	0.95	1.19	1.18	1.39	1.06	0.09	1.00
ESTIA total	1.02	0.91	0.92	0.83	0.97	1.40	1.00
Total	1.00	1.00	1.00	1.00	1.00	1.00	1.00

Source: UNHCR-ESTIA, 2018; UNHCR SMS, 2018; data processed by the authors.

), we use the location quotient (LQ) to demonstrate how the individual sites (Lavrio, Schisto, Skaramagas, and Eleonas) differ from the general norm (Total). It is important to note that individually they differ even though in total (Sites’ total), their composition does not differ from the Estia project (ESTIA total). We demonstrate how the ethnic composition of the sites differs from the overall composition and therefore one could argue that a pre-decided ethnic division is imposed. The share of Syrians in Schisto (15%) or Eleonas (29%) is much lower than the overall share of Syrians in the migrant population (44%). On the contrary, their share in Lavrio (54%) or Skaramangas (56%) is much higher. Respective differentiations are observed between the Afghan nationals in Schisto (74%)—often mentioned as the ‘Afghan village’—or Iranians (9%) and Pakistanis (5%).

The conservative government of New Democracy elected in 2019 introduced a stringent migration policy, and many reforms affected the operation of structures for asylum seekers and refugees. In 2020, uniform regulations for sites were established for the first time by a legal act and in 2021 a new law tightened the procedures for deportations and returning of third-country nationals. At the end of 2022, a new legal act launched Closed Controlled Structures on the islands and converted Open Sites to Controlled Temporary Accommodation Structures, imposing many security restrictions on the everyday lives of accommodated persons and restraining linkages to their environments. In the same period of 2022, the operation of the Elaionas site was discontinued as part of the regeneration plan of the area, which is one of the largest in the metropolitan region. The majority of the accommodated migrants were transferred to other sites in remote areas. The conservative mayor of the city of Athens supported the governmental plans despite the suggestions by the municipal opposition and humanitarian activists to include urban accommodation schemes in the regeneration plan and to provide support to the asylum seekers who evacuated the premises.

3.3. Vertical Segregation in Arrival Neighbourhoods

A first check of the profile of the URANUs where the ESTIA residences are located verified the assumption that EVS and the accommodation of displaced migrants are linked. Indeed, out of 713 URANUs accommodating ESTIA beneficiaries, 472 (66.2%) are found in high-rise areas. Furthermore, close to half of the 1134 high-rise URANUs of the metropolitan area (41.6%) accommodate ESTIA beneficiaries. This correlation becomes more evident for the areas where beneficiaries are over-represented, exceeding their average presence. With an average of 13.2 persons per URANU, we can locate 76 cases with more than 29 individuals (mean + 1 stdv), 72 of which (94.7%) are high rises susceptible to vertical segregation. Our next step was to identify the EVS profile of these areas.

Figure 3 depicts the results of the EVS analysis. We identified zones of vertical segregation (in green, red, and blue) as well as zones with no significant traces of vertical segregation (beige). Green areas indicate zones of vertical segregation where advantage and disadvantage coexist on a high level. Blue areas indicate zones of advantage for Greeks and those from developed countries located in the eastern and suburban parts of the high-rise apartment stock. Lastly, in the red colour, we identify zones of vertical disadvantage for foreigners from developing countries mostly located in the central and western parts of the city centre. While areas of vertical disadvantage (20.8%) show a high concentration, in reality, they are outnumbered by those of vertical advantage (21.7%), which are extremely scattered in inner suburban areas.

Figure 4 shows the distribution of URANUs according to the presence of ESTIA beneficiaries. Using a location quotient (LQ), we mapped their share to the permanent population (2011) in a given URANU, comparing it to their share in the study area. A LQ score of 1.00 indicates a presence similar to the average profile (1% of the permanent residents). The darker the colour, the greater the distance to the average with red nuances indicating over-representation and blue under-representation. The analysis reveals a stronger presence of ESTIA beneficiaries in the central parts of the city as well as in the adjacent municipalities of the western part of the metropolitan area. Many units of lower presence are also found in the western more remote low-rise parts of the city centre.

Areas with an under-representation of ESTIA beneficiaries (i.e., below the average 1% of the permanent population) compose 68% of the total number of URANUs. This confirms that the project, given the complexity of the mission undertaken, managed a high number of dispersions of beneficiaries. At the same time, the supply of properties that homeowners were willing to dispose of may better explain why areas with an over-representation of ESTIA beneficiaries are found in the disadvantaged parts of the city.

Figure 5 illustrates the spatial relationship between the data depicted in Figure 3 and Figure 4. The columns show how the groups of URANUs according to the presence of ESTIA-accommodated migrants (Figure 3) are distributed to different areas of vertical segregation (across the typology of EVS). Furthermore, the width of each column indicates the size of each group. We observe first that the higher the presence of migrants, the lower the possibility of a unit being found in areas non-susceptible to vertical segregation (in grey). The same goes for their presence in zones of limited vertical segregation (in yellow). Furthermore, their presence appears to have limited association with the areas of high and relatively high vertical segregation (in green) or of vertical advantage. This finding suggests a strong relationship between the presence of migrants and vertical disadvantage. In fact, 71.1% of URANUs of a higher over-representation of ESTIA beneficiaries are also areas of vertical disadvantage.

3.4. The Shaping of a Services Network and Urban Social Infrastructures

During our primary research, 335 services provided by 23 organizations were recorded, verified, and mapped (Figure 6). The operation of the ESTIA program, within the urban fabric, spurred the provision of supporting services—among others, training and support (29.6%), education (24.5%), employment services (9.6%), legal services (7.8%)—by NGOs and local government bodies to asylum seekers enrolled in the program and also to a wider population in a precarious living situation. It is noteworthy that social facilities were developed in arrival neighbourhoods and not close to camps (whereby basic services were also haphazardly offered by humanitarian organizations within the premises). Therefore, it does not come as a surprise that their spatial distribution (Figure 6) follows a similar pattern to the accommodation program, being over-represented around ethnically mixed and highly disadvantaged areas. From workshops with providers and policy makers, it emerged that this was a purposeful strategy initiated by larger NGOs providing both accommodation and services; it enabled them to serve the program beneficiaries within a feasible reach from accommodation units, manage the accommodation units with lower costs of supervision, and use the same buildings for housing their headquarters and specialised service facilities.

The concentration of services In proximity to central accommodation locations where the majority of migrants were accommodated was intending to address the challenges, reported in local studies, and especially the dehumanising conditions in camps [66,67,68,69] to enable access to integrated interventions and to improve coordination between civil and local government agencies. Our informants also emphasised why central locations facilitated intermediation to access specialised health services, which was important for vulnerable individuals and families, to address discrimination by housing landlords, and to enhance interaction with school and local communities. According to our informants, the provision of services in central locations led to an increase in the everyday mobility of migrants and visits to services, regardless of their place of residence, and this has also been reported in local studies [37]. This has in part compensated for the limited provision of services available in the port of Piraeus and in western suburban areas (Figure 6). In due course, apart from the initial providers, many other smaller charities and solidarity organizations, from below, were attracted, gradually increasing the density and the expansion of the services network to adjacent areas.

Figure 6 depicts this process of expansion. Dark green is the areas where the headquarters of organizations were used to offer their services within small shelters or in proximity to apartments they ran. Light green denotes the concentration of usually smaller organizations collaborating with accommodation providers for the provision of services. Reflections on this process by participants in workshops brought to the forefront how the pragmatic attitude of local authorities converged with the humanitarian aspirations of civil organizations, which were attracted to the program, and so from a theoretical point of view, the expansion and linking of services in central areas was vital for setting a solidarity agenda in local policies similarly to other European initiatives [26]. Furthermore, centrality enabled claims for a right to the city, squatting, and experimentation with communing practices of habiting by autonomous collectivities [70].

4. Discussion

Two types of arrival infrastructures have been identified in the preceding sections of this paper, each exhibiting not only a distinctive response to the mobility and establishment of migrant newcomers but also locations in areas that experienced and responded to the effects of the economic crisis differently. It is worth reflecting upon contextual factors, convergent and divergent strategies of actors involved in the formation of arrival infrastructures, and their contribution to the change in those peripheral and central city landscapes.

First, multilevel arrangements involved the UNHCR, the national and local authorities, and international and local civil society organizations. However, the role of the state and the national government needs to be emphasised for setting the preference over integration or control priorities when negotiating with the UNHCR, local authorities, and NGOs. This has become evident with the succession of the centre-left government of SYRIZA with an inclusive migration agenda, by the conservative government of New Democracy with a stricter control agenda.

Infrastructure of forced arrival established by the state, during the SYRIZA administration, played a decisive role with minimal and ex-post involvement of local authorities leaving non-governmental actors space to act within the premises of camps and responsibilities for establishing links to the urban environments. Not only were camps used as internal borderlands to contain an unwanted–hardly manageable population, but in terms of urban planning, they were treated as transient locations themselves—land reserved for unwanted uses until exclusive regeneration occurs. The conservative government of New Democracy discontinued the operation of many camps and imposed stringent rules on the operation of the remaining ones. Moreover, in operation today are sites located in areas for which no regeneration plans have been made. Regeneration by large investors is in progress in the areas of discontinued camp operations (Elaionas, Elliniko, Pireuas, Lavrio) without any inclusive plans for the accommodation of asylum seekers or refugees. Infrastructures of forced arrival could then be seen as parts of the ‘Operational landscapes’ of planetary urbanization, or ‘liminal landscapes’ [33], both containing wasted lives in dumping grounds, and anticipating urban explosion [31] and the transformation of rural and coastal peripheries, industrial and working-class suburbs, to landscapes of leisure and consumption [31,71].

Second, the partnerships and novel institutional arrangement between international and local agents of the civil society, municipal enterprises, and private landlords [66] were crucial for the formation of urban infrastructures in arrival neighbourhoods. In some respects, they became brokers of urban differentiation by managing the inherited built environment and by attempting to bring the private spaces of housing properties and dwelling under new institutional arrangements and linking them to social provisions. Nonetheless, this process was contradictory and ambitions for the expansion of inclusive spaces remained unfulfilled as long as the conservative government withdrew its support from urban accommodation schemes and evacuated buildings ‘occupied by illegal migrants’ [15,70]. Urban accommodation sustained an ephemeral regeneration of the rental market with affordable prices and upgrading of a decaying and vacant stock during the hardest years of the economic recession, which affected small private owners. Due to insufficient funding, the expansion of services was short-lived and progressively privatised. Moreover, as recovery progressed and after having renovated most of the premises used as accommodation units, private landlords opted for reusing or selling for homeownership or for Airbnb lettings. Given the hostile policy deterring the settlement of migrants in central areas, a private market of short-term rentals and homeownership of gentrifiers is expected to dominate. At the same time, a residual and informal housing market for the marginalised and displaced populations is most likely to consolidate.

5. Conclusions

Τhis article is one of the few recent studies on the impact of the main forms of arrival infrastructures on ethnic segregation (e.g., Gerten et al. (2023), Marcińczak and Bernt (2021) in Germany [27,28]). Whilst there is bourgeoning literature of the urban governance of forced migration in European cities, there are only limited studies that use a refined spatial analysis on arrival areas. Moreover, the majority of the urban governance studies focus on processes of integration rather than on processes of urban transformation. This article in part fills this gap and highlights the significance for combining the segregation literature with the investigation of urban policies for the development of reception infrastructures because their dispersion or concentration is shaped not only by ideological or political orientations but also by economic aspirations, real estate, and housing market functions.

The use of various data sources and segregation indices has enabled this article to quantitatively confirm the findings of previous qualitative urban policy studies in Greek and other European cities that camps contribute to increasing segregation in contrast to urban accommodation schemes.

Moreover, the spatial analysis has shown that the concentration of urban accommodation is double-edged. On the one hand, it is related to vertical segregation and to the disadvantaged placement of newcomers in the existing stock. On the other hand, it is linked to the actions of civil society organizations, which tend to concentrate in central city areas and have greater capacities for integrating housing with support services. This finding adds to the international comparison of arrival areas [27,28,30] and further suggests that the analysis of ‘micro-segregation’ can contribute to our understanding on how the built environment and local housing systems interact and shape opportunities for ethnically diverse and deprived inhabitants to access urban functions.

The theoretical framing of this research, posing questions and interpreting results, contributes to a multilevel, dynamic, and critical understanding of segregation. Inspired by Lefebvre [31] (pp. 77–103), it acknowledges the mediating role of the ‘urban’ level in responding to ‘global’ forces and introducing changes affecting the ‘habiting’ and every-day interactions. Ambrosini [17] intuitively suggested that local policies for asylum in Italian cities are a ‘battle-ground’ involving diverse actors; our framing expands his analytical metaphor to suggest that the transformation of arrival areas is a terrain for competing spatial strategies. Although these strategies can be discerned from their different orientation and ultimate objectives, they are dynamically formed and so they can better be understood as transformation paths rather than planning models [31] (pp. 135–150). Indeed, transformations are taking place concomitantly in distant and sparsely populated areas where it is easier to suspend the productive use of a surplus population, and in central decaying environments where habiting as a social need and human right is undermined by the conversion of dwellings into profitable machines.

The qualitative aspects of our research have highlighted that urban centrality offered opportunities for shaping a solidarity agenda where the strategies between the municipal authorities and international and civil organizations converged as has been noted in [26]. This agenda has temporarily contributed to hundreds of projects aiming to link housing with health, education, training, and employment services, and so our identification of a centralised network of infrastructures can be seen as a formalised and circumscribed vision of Lefebvre’s (1996/1968) classic concept of the right to the city. Centrality has also facilitated the multiplication of common practices in housing and health care [70], which neighboured or ad hoc cooperated with the formal structures, and so the city centre became a convivial place, assembling diverse practices for habiting and aspirations for change.

However, the humanitarian aspirations of our informants have not been materialised and indeed innovations have been halted by the imposition of the anti-migrant agenda by the conservative Greek government. This finding reaffirms the significance of the global level and the role of nation states in the international discussion of multilevel arrangements; it exposes the limits of local solidarity policies and the potentials of urban partnerships for addressing displacement [17,26,72]. Moreover, a southern European perspective highlights that real estate opportunities for small and financially strained homeowners or landlords superseded humanitarian concerns at a crucial timing when the institutionalization of housing innovations for vulnerable groups was instead necessary. It was also a lost opportunity for expanding innovations to address the experiences of precariousness in the everyday life of migrants, to enhance housing and mobility choices, to include their own communities in policy making, to establish service nodes in outer areas, and to enhance porosity along and across the edges of segregated city territories, to mention only a few of the suggestions we have recorded.

A limitation of our research results from using data collected at different time-points. This limitation can be addressed in the near future when uniform small-scale data from censuses will be available for many European cities. Thus, comparative research may explore in detail the changing urban geography of forced migration, and the role of arrival areas in the entry-ports of Europe and across the migration routes. Particular interest for comparisons lies in the systematic mapping of transformations in inner-city, suburban, and peri-urban areas and in the investigation of the concentrated spaces and expansive urbanization [32] resulting in segregation, and intensifying social inequalities. In this direction, availability of data from international organizations involved in the management and monitoring of asylum seekers and refugees can be an additional and valuable source.