The Regional Allocation of Asylum Seekers in Greece: A Multiple Criteria Decision Analysis Approach

Abstract

One of the long-term challenges for policy makers in host countries of migrants is the optimal geographical allocation of the migrant population so as to strengthen integration outcomes and serve the crucial goal of social inclusion. The political debate on the appropriate placement policy of newcomers has continued for years after the large-scale inflows of asylum seekers in Greece. This paper focuses on the evaluation of the dispersal policy of asylum seekers in Greece as implemented under the reception and accommodation scheme. Furthermore, it provides decision makers with an alternative dispersal policy framework driven by the integration outcomes of the already established migrant population in the country. The research encompasses Eurostat NUTS 2 annual data for a choice upon availability of the Zaragoza Integration Indicators and the immigrant population indicator for the thirteen Greek regions. The data cover a ten-year period from 2010 to 2019, and they are analysed with the multiple criteria decision-making method PROMETHEE. In addition to desk research, fieldwork has been conducted to illuminate the results of the model. The paper extends the literature on the governance of migration and contributes to the planning of migrant integration policies in their host countries.

1. Introduction

Europe’s biggest influx of humanitarian migrants after World War II [1] found Greece on the frontline facing unprecedented challenges. Greece did not share a long-term history in hosting asylum seekers. Thus, the national and local authorities were unprepared to respond quickly and effectively to the asymmetrical pressure of the large-scale arrivals. Having received more than 250,000 applications for asylum since the outbreak of the crisis [2], Greek authorities have already come to terms with two subsequent urgent issues: the allocation of these populations after first reception and an effective integration strategy for newcomers.

As Christopoulos [3] explains, the refugee issue is not a problem that can be solved but a phenomenon that should be managed, placing emphasis on the necessity of social integration. According to the Common Basic Principles for Immigrant Integration Policy in the EU [4], “Integration is a dynamic, two-way process of mutual accommodation by all immigrants and residents of Member States” and requires access to housing, employment, education, and social services. In the absence of integration measures or as result of their inadequate implementation, social benefits and funds could lead to institutionalization [3]. The New National Strategy for Integration in Greece refers to social integration as a two-way process of cooperation and dialogue between the hosting community and the immigrant population. Actions such as support for independent living, social housing, labour market integration, and social support have been among the priorities of the new strategy [5]. All the aspects of integration (economic, social, cultural, and political integration) are actually included in the concept of spatial integration [6]. That is, the residential concentration of migrants affects and is widely affected by variables such as housing, employment, social networks, etc. Thus, an optimal geographical distribution of the migrant population could play a crucial role in its integration outcomes. Nevertheless, in Greece, the political debate on the appropriate placement policy of asylum seekers arriving in the country has continued for years after the influx of 2015.

The aim of this paper is to define the optimal dispersion of asylum seekers in Greece in order to strengthen their integration outcomes and serve the crucial goal of social inclusion. The first objective of this research is to provide an allocation scenario for asylum seekers in the thirteen Greek regions annually, driven by the integration indicators of the already established migrant households in the regions. To this purpose, this paper uses available longitudinal data on the Zaragoza Integration Indicators from 2010 to 2019 processed with the well-known multiple criteria decision method PROMETHEE. Consequently, the model represents an ideal allocation pattern driven by integration criteria. Moreover, under this lens, the authors evaluate the regional dispersal policy of asylum seekers in Greece on the basis of the reception and accommodation scheme implemented in the country. Evaluation is based both on desk research and fieldwork. Fieldwork, mostly interviews with respective government authorities and other relevant stakeholders (International Organizations, NGOs, civil society actors), was implemented in the framework of the H2020 project “MigrAtion Governance and asYlum Crises (MAGYC)”.

The paper is structured as follows: the next section establishes the linkages between the importance of migrants’ integration and their residential placement, and it explores the relevant policies implemented in other EU countries. The structure with regard to the accommodation scheme for the newly arrived migrants and refugees in Greece follows. Upon this, the construction of the research model and the methodological approach are illustrated. Next, the results of the multiple criteria decision-making analysis as well as the outcomes of the fieldwork are presented. The conclusions are summed up in the last part of the paper.

2. Theoretical and Empirical Background

In general, EU migration policy has received much criticism. EU border governance has been considered as excessively violent [7]. According to De Genova [8], the crisis of border control and migration management in the EU is a crisis of sovereignty triggered by the autonomous subjectivity of human mobility and signals a crisis of the European social order. EU redistribution efforts in the field of migration and integration policy failed, illuminating the strong competence of the Union in regulation rather than redistribution, the lack of solidarity and the absence of centralized institutions at the root of such issues [9].

Dispersal policies lie at the heart of migration policies and management. The allocation policies of migrants in general, and asylum seekers and refugees in particular, include processes from their initial reception in a reception and identification centre and their transfer to another accommodation facility to the ultimate independent settlement in the host country, either following a relevant dispersal policy or an autonomous choice. Since it is probable that the migrants with a residence permit continue to stay in the area of their initial assignment, the accommodation scheme implemented by a host country includes several challenges.

The residential context has been used in the literature to explain the socioeconomic development of people. The residential environment is considered as an opportunity structure, a structuring context of social interactions or producing effects through interactions with external environments [10]. Therefore, the allocation mechanism for refugees could be seen as one of the first measures of integration policy [11]. Arrival areas and infrastructure represent a dynamic variable with significant impact on the integration of newcomers [12]. The geographical context, the personal characteristics, and the synergies between the two play a critical role in refugees’ integration outcomes [13]. Theoretical concepts describing residential segregation illustrate the importance of specific characteristics that could ease the transition and incentivize migrants to concentrate in specific residential areas. In particular, the spatial assimilation perspective describes the ethnic congregation of migrants in a host country so that they can receive help to access the labour market, housing, and education [14]. Ager and Strang [15] suggest that access to employment, housing, health, education, citizenship, rights practices, social connection processes, and the overcoming of structural barriers are key domains of social integration. The settlement and integration nexus involves the satisfaction of migrants’ personal needs such as housing, their access to employment, and their civic participation [16]. The importance of housing for social integration is also stressed in the EU Action Plan on Integration and Inclusion 2021–2027 [17] targeting the provision of adequate and affordable housing for migrants including social housing, the reduction of residential segregation, and the promotion of autonomous living. It should be mentioned, however, that the current management of asylum in the EU has been criticized for favouring repression rather than social integration [18]. The appropriate housing conditions are crucially important for the employment prospects, educational options, and social interactions of people in general and migrants in particular [19]. Thus, the settlement of migrants in areas with cheap available housing but low labour market demand is not a desirable solution to the dilemma of the optimal allocation of newcomers [20]. Employment enhances economic independence but also eases the social integration of newcomers [21]. However, housing shortages usually prevent policy makers from an optimal allocation of the migrant population based on the proximity of available job vacancies. Moreover, inadequate living conditions, such as the lack of accommodation, the poor conditions, and overcrowding, especially at reception centres, deter people from preparing to join the labour market. In the case when governments face the option of a fair spatial allocation between regions or an optimal cost minimizing allocation process, it is highly significant that they take into account the integration outcomes of migrants that are relevant to housing conditions [22].

There are various schemes suggested in the previous literature for the spatial allocation of the newcomer migrant population, and refugees in particular, based on algorithms that match efficiency with the preferences of both the interested parties: refugees and locals [23,24,25]. There is also the data-driven algorithmic assignment of Bansak et al. [13] that uses supervised machine learning and optimal matching to leverage synergies between the personal characteristics of the refugees and the resettlement sites to optimize integration outcomes. Hernes et al. [26] analysed the different settlement outcomes in Scandinavian countries, illuminating the conflict between dispersal settlement policies and favourable labour market integration outcomes. The European Council on Refugees and Exile [27] has suggested that qualitative and quantitative criteria, such as the GDP, size of the population, and ethnic networks, should be taken into careful consideration when a matching process between Member States and asylum seekers is taking place. Moreover, the report of the Council makes specific reference to the importance of access to education and the labour market for an effective integration outcome. Zimmermann [28] made reference to the criteria used in point systems to regulate migration, also including integration indicators such as unemployment rates. Aslund [29] studied the factors that influence individual location decisions and concluded that migrants prefer highly populated regions with large migrant and co-ethnic shares and available labour market opportunities. Marten et al. [30] have also paid attention to the role of ethnic networks to the economic integration of refugees. However, to the best of the authors’ knowledge, there is no research study employing the Zaragoza Integration Indicators either to identify an optimal allocation scenario for migrants, refugees, or asylum seekers based on integration outcomes or to evaluate migrant dispersal policies.

Most European countries follow a specific accommodation plan for the settlement of migrants, asylum seekers, and refugees, as reported in the relevant literature. Phillips [31] referred to the accommodation schemes for migrants in 15 member states of the European Union before the enlargement of 2004, identifying three integration patterns with distinct housing policy implications. Kearns and Whitley [32] underlined the UK’s dispersal policy debate, which has received attention for not contributing to the integration prospects of the immigrants. Korac [33] documented the importance of the conditions of settlement for the integration of refugees and the negative effects asylum centres had in the Dutch case. Dispersal policies followed by many countries, such as Denmark, Germany, the Netherlands, Sweden, and the UK, have been blamed for putting the hosting cities in a position of containing individuals and acting as backdrops to political decisions [34]. As regards the allocation scheme of asylum seekers in Greece, most prominent reasons for the dispersal policy followed could be the concerns of the local communities and/or local capacity constraints, as indicated through the interviews implemented by the authors. The willingness of the community to host the migrant population is actually a factor taken into consideration while planning and implementing the dispersal policies in other European countries, such as Finland and Portugal [20]. Overall, effective social inclusion requires that settlement policies are part of wider integration strategies [35].

The importance of establishing certain inclusion criteria when designing dispersal policies is demonstrated in the OECD’s booklet on integration [20]. The booklet presents the dispersal criteria of asylum seekers and humanitarian migrants in OECD countries as derived from a query of the European Migration Network [36] and an OECD questionnaire on the integration of humanitarian migrants [37]. Greece is not among the countries following a deliberate dispersal policy for asylum seekers with certain criteria. After all, the Southern European countries do not share a tradition in welfare provisions for migrants and rely on informal channels rather than on public interventions to promote residential integration [31]. On the other hand, Scandinavian countries, such as Finland and Sweden, have incorporated employment opportunities in their dispersal schemes. Poland has a specific point system of the dispersal criteria. Germany follows firm quotas to avoid minority ethnic segregation. Denmark also uses quotas in its planned settlement program for humanitarian migrants [20]. In Norway, the program is based on voluntary local-central agreements with selected municipalities, while in Sweden, there is individual autonomy through self-settlement [26]. In general, the lack of systematic data, especially in the case of humanitarian migrants, such as their preferences, their background characteristics, or even the difficulties in the recognition of their skills, may render the use of the aforementioned mechanisms in a wide range unavailable.

This paper focuses on taking advantage of the European toolbox as regards migrant integration data to highlight their value in migration governance and in the planning of migrant integration policies, paying particular attention in the role of labour market integration for the successful settlement of the migrant population.

3. Asylum Seekers’ Accommodation Scheme in Greece

One out of the two million people who entered the European Union since 2014 have used the Eastern Mediterranean Sea route [38]. In fact, in 2016, Greece became the first European country with the highest number of asylum applications compared to its population [39]. Initially, the focus of public action was largely on reception measures. Since many of the newcomers stayed in the country, a revised integration policy became an emergency issue. Greece, while receiving inflows of asylum seekers, was experiencing a recession, which deeply affected its labour market and the third-country nationals already integrated into it.

As of January 2020, there were more than 115,600 new migrants and refugees in Greece [40]. The accommodation scheme for the asylum seekers includes the Reception and Identification Centres (RICs) or the hotspots, the open temporary reception facilities, the facilities for asylum applicants [41], the UNHCR Emergency Support to Integration and Accommodation scheme—ESTIA, and the apartments rented by other schemes involving municipalities, NGOs, and the Orthodox Church of Greece.

This accommodation scheme has received a lot of criticism. Hotspots have especially been criticized as detention and foreclosure facilities that have been applied by the EU to deter new migratory flows [18]. In fact, Greece would be penalized by the EU unless hotspots were established within a deadline of three months [42]. As regards the ESTIA project, housing conditions are better than in the camps, but the social integration services offered have mostly been voluntary due to lack of funding. For example, Apostoli, an NGO of the Orthodox Church of Greece, is in cooperation with the UNHCR in the ESTIA framework to offer accommodation to unaccompanied minors and vulnerable people [43]. Kourachanis [44] concludes that ESTIA is a minority social intervention that lacks the holistic approach that the term “social integration” requires and the use of which legitimizes EU policies of repression and border control. Moreover, Kourachanis [44] stresses that defining vulnerable people as the target group of ESTIA social housing indirectly legitimizes residence in camps as tolerable with acceptable housing and health conditions. Problems such as overcrowding of the program, inexperienced staff, and burnout of the labour force have been faced in the ESTIA provision of services as well as in most of the migration professionals [18,45].

The main form of asylum seekers’ housing in Greece remains accommodation in camps. Again, this type of accommodation has been criticised widely in Europe. Szczepanikova [46] argued that accommodation centres, referring to various forms of collective housing such as camps, are a tool of migration control, which is closely related to the assistance offered to asylum seekers resulting in their dependency. Davies et al. [47] showed how both inactivity and political actions can become a means of coercion, control, and power and how the abandonment of refugees in makeshift camps inside Europe led to their exposure to violence, usually structural.

As of October 2020, there were 30 mainland long-term accommodation sites in Greece, one transit site, and one site with nine hotels. There were 5918 accommodation units in total with a capacity of more than 31,000 places [48]. The Migration and Asylum Ministry took administrative control of 28 of the migrant shelters with a Joint Ministerial Decision which was published on 24 March 2020 [49]. Ten sites were located in the region of Central Macedonia, seven of them in Attica, two in the region of Western Greece, four of them in Central Greece, three in Epirus, two in East Macedonia and Thrace, and two in Thessaly. There were also the RICs on the islands [48]. It was estimated that the facilities on the islands hosted 42,000 people in February 2020, despite their limited capacity of 5400 [50]. ESTIA, offering urban accommodation, has benefited more than 64,600 people since its launch in January 2015, with 4604 apartments and 8 buildings in 21 cities in Greece [40]. With regard to unaccompanied minors, it was estimated that 5232 were residing in Greece. Attica hosted 64 long-term and 4 temporary accommodation facilities for unaccompanied minors, and Central Macedonia hosted 10 and 8, accordingly [51].

The legislation with regard to the migration and asylum governance has undergone several amendments after the outburst of migrant inflows. As a result, the reception organization followed institutional initiatives and became rather complex, involving both state and non-state actors. Moreover, every amendment was followed by oppositions to the plans, such as those of the residents on the islands in 2020, when the Greek government passed a new reform which provided for the construction of new semi-closed detention centres on the islands [50].

Overall, Greece has been questioned for the suitability of housing conditions and decent infrastructure that meet every day needs available to asylum seekers, particularly highlighting overcrowding, health risks, and a lack of social care [18].

4. A Multiple Criteria Decision-Making Analysis

To explore the optimal allocation of asylum seekers in Greece, this paper uses the methodological approach based on the Preference Ranking Organization METHod for Enriched Evaluation. The PROMETHEE method was first launched as PROMETHEE I (partial ranking) and PROMETHEE II (global ranking) in 1982 by Jean-Pierre Brands, while later in 1995, PROMETHEE III and PROMETHEE VI were developed by Jean-Pierre Brands along with Bertrand Mareschal. Services, public and industrial applications, energy and water management, economics, and transportation problems are among PROMETHEE’s main application areas.

PROMETHEE is a method that belongs to the outranking family of multi-criteria methods and is based on pairwise comparisons in order to calculate the preference degrees. The PROMETHEE II method provides the decision maker with a final ranking of the selected alternatives. The ranking will be based on the computation of preference degrees and the decision maker will have a single choice or a choice over alternatives as to the best solution of the set group of alternatives and their preference degrees. There are several steps towards producing the final ranking of the selected alternatives following the PROMETHEE method. The first step includes constructing the table of the decision matrix, integrating the alternatives, criteria, preference parameters, and weights. The next step refers to the calculation of the differences between the actions’ evaluations for each criterion and the pairwise comparison matrix for each criterion. Then, the method calculates the unicriterion net flows, the weighted unicriterion flows, and the global preference net flows. In the end, the result is the ranking of the actions according to PROMETHEE II [52,53,54].

The criteria used in the method could be qualitative as well as quantitative. The method also requires as inputs the weights of the selected criteria, the preference function’s type (Usual, U-shape, V-shape, Level, Linear, and Gaussian), and the appropriate thresholds. This research model applies the Usual function for all the criteria except for that of the foreign-born population. The Usual function does not require an indifference threshold q and a preference threshold p to define the preferences of the decision maker. Nevertheless, the Linear function was applied to the size of the foreign-born population criterion, specifying 5000 as an indifference threshold q and 20,000 as a preference threshold p. In particular, when the difference between a criterion’s evaluations is less than the selected threshold of indifference q, there is no perceived discrepancy between these two actions for the preferences of the decision maker. On the contrary, when the difference is higher than the preference threshold p, then the preference is strong [54].

For the purpose of this paper, a model scenario with the 13 Greek regions as the alternatives and a selection upon availability of the Zaragoza Integration Indicators as reference criteria has been constructed. The Zaragoza Declaration adopted in April 2010 [55] and approved at the Justice and Home Affairs Council of the EU on 3–4 June 2010 defined the EU integration indicators in an effort to harmonize the assessment of the migrants’ integration policies’ outcomes between the EU member states. The indicators help understand, evaluate, and compare the national integration policies. The Zaragoza Indicators include twenty-one indicators in five areas of interest, such as employment, education, social inclusion, active citizenship, and welcoming society. Due to the unavailability of adequate longitudinal regional data on all the suggested indicators, the selection was based on the availability of annual statistical data by Eurostat [56] for the time period 2010–2019 for the 13 Greek regions following the NUTS 2 classification.

Hence, for the purpose of this paper the authors used NUTS 2 regional data for the employment rate, the unemployment rate, the activity rate, the number of the early school leavers, and the population of foreign-born people (expressed in thousands) as reference criteria. The indicator of the foreign-born population has not been included in the Zaragoza Integration Indicators list, but it was selected to serve as a balance to the model.

In detail, the employment rate expresses the percentage of employed people compared to the working-age population, while the unemployment rate refers to the percentage of the labour force that is unemployed during the reference year. The activity rate is an indication of the economically active people compared to the total population. Early school leaving has been represented in the sample by the statistical data on the population with less than primary, primary, and lower secondary education. Data on all the aforementioned indicators refer to the foreign-born population between 15 and 64 years old.

At this point, it should be mentioned that the source for the statistical information provided by Eurostat and formulating this model’s dataset is the EU Labour Force Survey, which is a quarterly household sample survey conducted in the EU member states covering the total population usually residing in Member States, except for persons living in collective or institutional households. Moreover, from this dataset, we have selected indicators that cover persons in private households aged 15–64 years old. Taking into consideration that (a) the major influx of asylum seekers in Greece took place after 2015, (b) there was a prolonged period of waiting for an asylum decision, and (c) the duration of accommodation in collective/institutional accommodation schemes was usually extended, it could be assumed that the statistical sample used in this model does not include the majority of the newcomer migrants in Greece. It rather uses the integration indicators of the already resident and established migrant and refugee population in the country as inputs so as to provide a pattern of allocation driven by migrant integration indicators. In this framework, the results of the PROMETHEE II method are analysed focusing on the integration outcomes of the already established migrant and refugee households in Greece that figure an allocation pattern for the new migrant inflows as well.

Table 1. PROMETHEE preferences.

Preferences	2010 to 2019
	Employment Rate	Unemployment Rate	Activity Rate	Less than Primary, Primary, and Lower Secondary Education (Levels 0–2)	Population
Min/max	max	min	max	min	max
Weight	1	1	1	1	1
Preferences function	Usual	Usual	Usual	Usual	Linear
Thresholds	absolute	absolute	absolute	absolute	absolute
Q: Indifference	n/a	n/a	n/a	n/a	5
P: Preference	n/a	n/a	n/a	n/a	20

demonstrates information on the preference parameters of the model. The employment and the activity rates should be maximized in the optimal case of social inclusion of the migrant population, while the unemployment rate and early school leaving should be minimized. The population criterion has been set to be maximized in order to serve as a balance to the model as well. All the criteria are quantitative. The weights of all the criteria are equal because there is no question of difference of importance among the criteria.

Following the above-mentioned parameters,

Table 2. PROMETHEE results.

	2010	2011	2012	2013	2014	2015	2016	2017	2018	2019
1	Attiki	Attiki	Peloponnisos	Attiki	Attiki	Attiki	Attiki	Kriti	Voreio Aigaio	Kriti
2	Peloponnisos	Sterea Ellada	Sterea Ellada	Kriti	Notio Aigaio	Peloponnisos	Kentriki Makedonia	Attiki	Attiki	Attiki
3	Kriti	Peloponnisos	Attiki	Peloponnisos	Kriti	Kentriki Makedonia	Voreio Aigaio	Dytiki Ellada	Kriti	Notio Aigaio
4	Anatoliki Makedonia, Thraki	Kriti	Kriti	Thessalia	Dytiki Ellada	Kriti	Anatoliki Makedonia, Thraki	Voreio Aigaio	Notio Aigaio	Voreio Aigaio
5	Voreio Aigaio	Voreio Aigaio	Ionia Nisia	Notio Aigaio	Peloponnisos	Anatoliki Makedonia, Thraki	Kriti	Peloponnisos	Peloponnisos	Anatoliki Makedonia, Thraki
6	Sterea Ellada	Kentriki Makedonia	Notio Aigaio	Kentriki Makedonia	Ionia Nisia	Ionia Nisia	Dytiki Ellada	Sterea Ellada	Sterea Ellada	Peloponnisos
7	Kentriki Makedonia	Thessalia	Kentriki Makedonia	Sterea Ellada	Kentriki Makedonia	Sterea Ellada	Ionia Nisia	Kentriki Makedonia	Kentriki Makedonia	Kentriki Makedonia
8	Dytiki Ellada	Anatoliki Makedonia, Thraki	Ipeiros	Ionia Nisia	Thessalia	Notio Aigaio	Sterea Ellada	Ionia Nisia	Anatoliki Makedonia, Thraki	Sterea Ellada
9	Ionia Nisia	Ionia Nisia	Voreio Aigaio	Voreio Aigaio	Anatoliki Makedonia, Thraki	Dytiki Ellada	Peloponnisos	Notio Aigaio	Dytiki Ellada	Dytiki Ellada
10	Notio Aigaio	Notio Aigaio	Anatoliki Makedonia, Thraki	Dytiki Ellada	Sterea Ellada	Voreio Aigaio	Notio Aigaio	Thessalia	Ionia Nisia	Ionia Nisia
11	Thessalia	Dytiki Ellada	Thessalia	Anatoliki Makedonia, Thraki	Voreio Aigaio	Thessalia	Ipeiros	Anatoliki Makedonia, Thraki	Ipeiros	Thessalia
12	Ipeiros	Dytiki Makedonia	Dytiki Ellada	Ipeiros	Dytiki Makedonia	Ipeiros	Thessalia	Ipeiros	Thessalia	Dytiki Makedonia
13	Dytiki Makedonia	Ipeiros	Dytiki Makedonia	Dytiki Makedonia	Ipeiros	Dytiki Makedonia	Dytiki Makedonia	Dytiki Makedonia	Dytiki Makedonia	Ipeiros

includes the preference rankings for the decade 2010–2019. As depicted in the table, the region of Attiki (Attica) was ranked first for most of the years examined with regard to the integration outcomes of migrants and refugees, and consequently, it was the most preferable area to host the newcomers as well. This result was to be expected since the capital city of Greece, Athens, is located in Attiki (Region of Attica), so it should fulfil at least some of the criteria for the migrants’ integration compared to the other regions. In more detail, almost half of the country’s population is concentrated in Attiki (Region of Attica). As a result, there are more employment and housing opportunities and larger social networks to facilitate the integration process. Hence, there is proper argumentation to support the fact that 40% of the asylum seekers residing in the facilities under the accommodation scheme running in Greece were concentrated in this region.

On the other hand, the second most populated region in Greece, the region of Central Macedonia, accommodating more than 10,000 asylum seekers, held an average place in the ranking except for 2016, when it received the second highest rank among the thirteen regions. This could most probably be attributed to an increase in the migrant population in the region due to the closure of the Balkan corridor towards Europe. As a result, many migrants were stranded in the region bordered to the north by North Macedonia. At this point, it is worth mentioning that it was in 2016 when the mainland camps were created to decongest the islands and to host the asylum seekers. This was also the year when the informal makeshift camp of Idomeni, which captured the media’s interest for months, was populated.

The region of Kriti (Crete), which is the biggest island in Greece, held a high rank before and after the peak of the migrant inflows in 2015 according to the results. However, only a small portion of the asylum seekers was placed in accommodation sites on the island. On the contrary, Sterea Ellada (Region of Central Greece), which is the third most populated region with asylum seekers in institutional accommodation places, received an average evaluation relative to the other regions after 2015. However, it should be noted that Sterea Ellada (Region of Central Greece) shares internal borders with Attiki (Region of Attica), so it is easily reached from the capital city of Athens.

Proceeding further with the results, the region of Voreio Aigaio (North Aegean) was placed high in the ranking after 2015. The region is on the Eastern Mediterranean Sea route used by the irregular migration flows as the entrance point to Greece. Lesvos, Samos, and Chios, where three of the RICs were placed, belong to this region. The outcome of the comparative analysis placing this region high in the ranking could be accounted for by this feature alone. That is, apart from the thousands of people held in the RICs, there were many migrants residing independently on these islands, benefitting from the employment opportunities that occurred after the migration outbreak. Still, Anatoliki Makedonia-Thraki (Region of Eastern Macedonia and Thrace), including the land borderline of Greece with Turkey and one RIC and hosting almost 1500 asylum seekers, was ranked lower than the region of Voreio Aigaio (North Aegean) between 2016 and 2019.

What is more interesting, however, is that the region of Notio Aigaio (South Aegean), which hosted only 374 asylum seekers under the accommodation scheme, gradually improved its place in the ranking, favouring the distribution of a larger share in the regional allocation of asylum seekers. On the contrary, the regions of Ipeiros (Epirus) and Thessalia (Thessaly), which hosted more or less than 3000 newcomers, were among the least favourable regions for their allocation driven by migrant integration factors. The region of Peloponnisos (Peloponnese) in Southern Greece used to hold a high place in the ranking before 2013, and it used to be a common destination for labour migrants employed in the agricultural economic sector. However, it was downgraded from fifth to sixth place in the ranking in 2019.

Last but not least, Dytiki Makedonia (Region of Western Macedonia), being among the top regions in Greece with the highest unemployment rates during the last decade, received the lowest rank for most of the years examined. Dytiki Ellada (Region of Western Greece) was ranked ninth among the thirteen regions for 2018 and 2019 as projected in the table. Thus, it is reasonable that Dytiki Makedonia (Region of Western Macedonia) and Dytiki Ellada (Region of Western Greece) were not among the top destinations in the migrant distribution policy followed in Greece. Likewise, Ionia Nisia (Region of Ionian Islands), which had relatively low ranks after 2015, did not host any migrant accommodation site.

The aforementioned results presented in Table 2 were based on the scenario modelled by the authors focusing on specific indicators for a certain time period and under specific preferences. For this reason, a sensitivity analysis of the weights was conducted to examine the robustness of the ranking. In the case when the intervals of all the criteria are particularly limited, the given solution is characterized as quite sensitive to change, and the decision maker could consider an alternative model constructed to be more stable. On the contrary, when the stability intervals are wide, the suggested ranking indicates robustness.

Table 3. Sensitivity analysis.

Stability Intervals in %
	2010	2011	2012	2013	2014	2015	2016	2017	2018	2019
Employment rate	19.73–21.16	16.29–23.83	18.26–23.95	17.56–22.15	16.80–20.12	17.38–22.33	13.72–20.21	19.11–21.21	18.70–25.23	19.03–20.07
Unemployment rate	19.64–21.56	14.16–23.09	17.53–21.74	19.31–23.97	19.00–20.37	17.40–25.58	19.14–25.00	18.92–20.44	18.86–25.76	18.83–20.15
Activity rate	19.73–20.81	17.55–25.59	17.42–22.88	18.42–21.98	15.95–20.37	17.92–21.73	17.13–20.42	19.11–20.17	17.10–21.75	19.93–25.37
Less than primary. primary and lower secondary education	17.94–20.15	17.08–22.03	17.42–21.54	18.08–20.38	18.78–20.37	18.64–20.84	18.26–23.67	19.69–20.87	15.13–22.18	19.61–21.14
Population	18.33–20.27	15.12–22.75	14.89–22.05	18.01–25.08	19.94–23.43	15.96–26.86	19.81–24.41	19.60–21.95	15.26–21.32	19.62–21.22

illustrates the intervals for the selected criteria for the decade 2010–2019. Although it is clear that the limits above the intervals are not wide enough, the strict limits could be attributed to the fairly close values of the data entered into the model. As a result, the sensitivity of the results and the low flexibility of the model are to be expected, and they do not play a negative role in the given rankings. However, the analysis provides stimuli for further research with alternative models, using, for instance, different weights, in order to focus more on selected criteria and examine how the ranking will react.

5. Fieldwork Outcomes

The outcomes of the fieldwork conducted by the researchers make the aforementioned analysis more comprehensive. The research team conducted over 20 semi-structured interviews with stakeholders in summer 2020 in the regions of Attica, Central Macedonia, Crete, and Thessaly. Due to the COVID-19 pandemic restrictions, access to the migrant population and particularly those accommodated in the camps has not been accessible. As a result, the sample consists of stakeholders including NGOs, civil society organizations, international organizations, and government authorities active in the field of asylum seekers’ urban accommodation under the ESTIA project implemented in Greece by UNHCR [57]. The key questions focused on the division of responsibilities between competent bodies, the coordination of actions by responsibility (reception, accommodation, integration), the effectiveness of the activities and their impact on integration prospects, and the role that the attitude of the indigenous population plays.

In line with the interviews, our understanding is that the allocation of asylum seekers in Greece was a responsibility of the central government and did not follow a plan driven by specific criteria. The willingness of the local society to host migrants could be a probable factor that affected the dispersal decisions. The interviewees have not reported any significant reactions at the local level with regard to the function of the UNHCR urban accommodation scheme ESTIA. They did notice, however, that the camps were a whole different type of accommodation that usually raised concerns with the local communities. As regards the residents of the Greek islands which have been the main gates of the asylum seekers and host the RICs, they felt that they were carrying an uneven burden contrary to the rest of the country.

It should be mentioned at this point that at the beginning of migrant inflows, Greeks greeted the asylum seekers with open arms. It was soon enough that the consequences of the influx became obvious, and the oppositions started arising. After other European countries closed their borders, migrants were stranded in a country which was suffering from a deep recession. One of the first results of the situation was the reduction in the tourist arrivals on the islands, which was a vital part of their economic activities. Moreover, a large part of the concerns was focused on the perceptions of increased competition over the same job positions in a period of austerity that followed the economic crisis in Greece and led to low labour market demand [58]. Then, there again was the issue as regards the degradation of the environment due to the congestion in specific places, which lacked the necessary infrastructure to host them, such as an improved sanitary system and a waste management system. Thus, public health issues also arose [59]. The escalation of crime in the congested areas has bothered the authorities from the beginning of the crisis [60]. Last but not least, demographic changes occurred, such as those in the case of Moria (21,000 migrants and less than 3000 nationals), which increased the competition between the residents for social services such as access to healthcare [61]. Hence, local communities can play an important role in the implementation of dispersal policies.

With regard to the concentration of asylum seekers in urban areas, it was common belief among the interviewees that vulnerable migrant families specifically need to be allocated close to health services. However, urban areas also offer proximity to humanitarian organizations, employment and housing opportunities, and the existence of migrant networks. Although they acknowledged the value the newcomers could add to the primary sector of the Greek economy, they also mentioned that this could happen under several circumstances that have not been fulfilled so far in Greece. Measures such as language learning and vocational training as well as social activities where migrants and natives interact were suggested as important steps towards the rural settlement of asylum seekers. This could explain the position of the region of Thessaly in the ranking, which could be a potential host region for low-skilled migrants due to its advanced primary economic sector, but it does not meet the prerequisites mentioned above. In the case of independent living, the stakeholders reported that most of the migrant population searched to settle in places where there were migrant networks and employment opportunities, such as urban areas and specifically the region of Attica. With regard to the region of Crete, the stakeholders mentioned that the island offers employment opportunities in agriculture, livestock farming, and the tourism sector, and targeted allocation and integration measures could contribute to the inclusion of the new population. More innovative policies, such as the promotion of social cooperative enterprises, were suggested as necessary in areas deeply affected by the economic recession, such as the regional section of Kilkis in Central Macedonia. Overall, it could be concluded that urban centres have been more favourable for the accommodation of asylum seekers in Greece so far. Targeted integration measures could render the allocation in rural areas a win-win situation for the migrant population and the local society in the future.

Combining desk research with fieldwork outcomes was challenging since an important factor stressed by the interviewees as regards the distribution of asylum seekers (the attitudes of the local communities) was not included in the model scenario. The authors envisaged an asylum seekers’ allocation model scenario based on the available EU integration indicators, which have been under-used as referred to in the EU Action Plan on Integration and Inclusion 2021-27 [17] to support sustainable independent living and an inclusive society. Researchers could consider including a quantifiable indicator for the local reactions towards asylum seekers’ accommodation structures among the dispersal criteria in future studies and adjust their preference degrees so as to deliver an alternative asylum seekers’ allocation scenario that could serve the diverse commitments of policy makers. Finland and Portugal, as mentioned above, include the willingness of the local community to host migrants in their dispersal criteria. Future research could synthesize the EU toolbox with the elements engaged in national dispersal schemes to offer an integrated decision-making model.

6. Conclusions

Years after the outburst of migratory flows towards Greece, the appropriate dispersal policy for asylum seekers in the Greek regions has continued to be part of the public dialogue. Greece did not share a tradition in intervening in the placement of migrants and refugees. In fact, Greece was not included in the ad hoc query on monitoring integration requested by the European Migration Network on 21 December 2018 [62]. Forasmuch as the allocation of asylum seekers is part of the integration process and thousands of them leave Greek institutional housing as soon as they receive a positive decision on their asylum application, their efficient distribution following specific socio-economic features is crucial for their effective integration. In other words, a dispersal scheme based on well-founded indicators could improve the efficiency of the Greek migrant integration policy. There are examples of other EU Member States that follow a specific dispersal policy for the migrant population. In fact, some of them, such as Finland and Sweden, have incorporated integration indicators in their migrant allocation schemes.

This paper used the PROMETHEE multiple criteria analysis approach to provide a comparative ranking of the Greek regions based on a choice upon availability of the Zaragoza Integration Indicators that could be implemented in the case of other countries as well. The ranking offers a pattern for asylum seekers’ allocation in regions where the integration outcomes of the already established foreign population are more favourable. According to the results, following the first research question, the concentration of asylum seekers in accommodation facilities around the capital city is more preferable for helping them adapt to independent living. On the other hand, as regards the second research question of this paper, the reasoning supporting the placement of a considerable number of asylum seekers in regions such as Epirus and Thessaly could not be clarified based on their low position in the preference ranking. Fieldwork offered evidence as to the relevance of urban areas for migrant integration and the obstacles making migrant integration difficult at the moment in rural areas. Moreover, the interviews highlighted the role that the response of the local population played in the dispersal policy of asylum seekers determined by the Greek authorities.

Early action and the availability of integration data need to be improved in Greece and the other EU Member States to allow for evidence-based decision making on integration governance [17]. Indicators such as employment opportunities, available housing, and the availability of integration services should be taken into account in the planning and implementation of allocation policies [20]. As indicated in this paper, criteria already successfully implemented in national migrant dispersal schemes, such as the willingness (or not) of the local communities to host migrants, could offer further value to the decision-making models determining the dispersal policy in favour of long-term effective social inclusion of newcomers. To this end, EU Member States could contribute to the availability and accessibility of statistical data as regards the aforementioned indicators. EU countries with alternative dispersal schemes for the migrant population could take advantage of the EU toolbox to facilitate comparative analysis and promote mutual learning and the PROMETHEE method, suggested in this paper for the case of Greece, to adjust and evaluate their policies.

The decision-making model applied in this paper suggests an alternative approach to planning and formulating the asylum seekers’ allocation policy driven by the available EU accepted integration indicators in order to support sustainable migrant integration and sound migration governance and facilitate international comparisons. The extension of accessibility to statistical data that could offer migrant dispersal criteria, along with the flexibility that PROMETHEE provides decision makers to decide and change the preference degrees of these criteria, could be a valuable asset in the field of EU migrant allocation policy.