At Europe’s frontline: factors determining migrants decision making for onwards migration from Greece and Turkey

ABSTRACT

Limited research has been conducted on migrants’ decision making factors in transit and this is an important area of investigation that goes beyond the origin-destination country dichotomy most commonly represented in migrant decision making. This paper examines the decision making factors for onwards migration or stay of migrants from Afghanistan, Iran, Iraq, Pakistan, and Syria in Greece and Turkey. This paper is based on a unique dataset of surveys collected with 1056 migrants in Athens and Istanbul from May- July 2015. The decision for onwards migration or stay is examined through a probit regression analysis. Four categories of influencing variables are discussed: the country of origin, current conditions in the transit country, the initial migration destination choice and previous onwards migration attempts. The results demonstrate that first; the majority of respondents in both countries seek to migrate onwards (75% in Greece and 63% in Turkey) and that conditions in the transit country are highly significant in influencing onwards migrate decisions including their current subjective living conditions, employment, experiences of abuse, and speaking the local language. This paper contextualizes the findings and highlight the complexity of factors influencing migrants’ decision making in Greece and Turkey.

KEYWORDS:

• Transit
• decision making
• irregular migration
• Greece
• Turkey

Introduction

In 2015, Europe received an unprecedented number of arrivals with 857,363 migrants and asylum seekers arriving in Greece alone, virtually all of which came to Greece from Turkey and 853,650 arrived by sea (IOM, 2016a). Both Greece and Turkey have represented significant countries of destination and transit for migrants over the past decade. However, within the context of migration to Europe these countries have critically different roles with Turkey being the last point of departure for the European Union (EU) and Greece being the frontline of entry to the EU. The media representation of the migrant flows of 2015 in Europe have highlighted the role of both of these countries primarily as sites of transit and less as sites of destination. This paper examines the dual role of these countries as sites of transit and destination in the context of the 2015 flows and specifically assesses how migrants decide to stay in these countries or migrate onwards from them.

A substantive gap exists in the migration literature on migrants’ decision making factors in transit (Townsend & Oomen, 2015). There is increasing recognition that first, migrants’ decision making is influenced by an array of complex factors arising between the destination and origin country (De Haas, 2011; McAuliffe, 2013; Wissink, Duvell, & van Eerdewijk, 2013) and second, that migrant’s decision making needs to be understood across each stage of the migration journey (Koser & Kuschminder, 2015; Townsend & Oomen, 2015; Wissink et al., 2013). Decision-making in transit is thus a critical area for exploration as it goes beyond the destination-origin dichotomies frequently used in the migration debate and it represents a formidable aspect of migrants’ journeys.

The objective of this paper is to examine the factors that influence migrants’ decision making to either stay in a country of transit or migrate onwards. The two case studies used in this paper are Greece and Turkey, both of which have been at the frontline of migration to Europe in 2015 and represent significant countries of origin, destination and transit migration. The paper is based on an original dataset of 1056 migrants from Afghanistan, Iran, Iraq, Pakistan and Syria collected from May-July 2015 in Athens and Istanbul. The data was collected by using computer assisted personal interviewing (CAPI) on tablets with a structured life-cycle questionnaire. In order to understand the migration decision, we examine four categories of influencing variables: conditions in the country of origin, current conditions in the transit country, the initial migration destination choice and previous onwards migration attempts. It is important to stress that this paper focuses on the context of Greece and Turkey up to the end of July 2015. From August 2015, the environment began to rapidly change with events such as Angela Merkel’s welcoming of Syrian refugees, followed by the opening and closing of the Balkan route from Greece in the subsequent months. All of these events may impact the decision making of migrants and cannot be captured or discussed within the scope of this paper.

This paper is divided into seven following sections. The first section of this paper seeks to contextualize the concept of transit and the second section provides an overview of the current literature on migrant decision making processes in transit. The third section gives a brief introduction to the two case studies of Greece and Turkey and the forth section presents the approach that we use to measure decision making factors in transit. The fifth section presents the data and methods used in the study, the sixth section the results and finally the discussion and conclusion.

Greece and Turkey as spaces of ‘transit’

The term ‘transit migration’ and ‘transit countries’ emerged in migration debates in the 1990s in particular in reaction to changing patterns of migration to Europe (Collyer & de Haas, 2012). Despite the increasing prevalence and use of the term, there is no agreed upon definition of transit migration (Düvell, 2012). Key unresolved issues in the definition include: intentions of the migrant, such as the intention to migrate onwards or not; duration of the migration, in other words when is a migrant no longer considered to be in transit; and the relationship between legal status and transit (Düvell, 2008).

It is important to note that transit migration is a term that is often associated with forced migration and irregular migration (Hugo, Tan, & Napitupulu, 2014). Limited options for direct legal migration to destination countries, it is argued, pushes people into irregular migration movements that are often facilitated by smugglers and include long journeys through multiple transit countries (Düvell, 2014). The result has been a commonly negative perception of transit migrants; which may explain challenges such as discrimination that many face en route.

Further to this, there is increasing argumentation to move away from the term ‘transit migrant’. For example, Yildez and Sert (2019), argue for the term mobistasis to reframe transit as a point of hiatus and waiting during the asylum journey. Taken from a journey perspective (Benezer & Zetter, 2015), the unit of analysis would be a moment within a journey. Finally, from a trajectories approach (Schapendonk, van Liempt, Schwarz, & Steel, 2018) the analysis would focus on moving through a place and the ways in which open spatio-temporal processes occur. Some of this literature directly contrasts the notion of the transit (Yildez and Sert, 2019), whereas others use the notion of transit, but recognize its limitations as still suggesting a linear approach (Benezer & Zetter, 2015; Schapendonk et al., 2018). In this paper, we recognize these limitations of the term and do not refer to migrants themselves as ‘transit migrants’.

From an EU policy perspective, the term ‘transit country’ has typically referred to the countries at the EU’s external borders, such as Morocco, Turkey, Libya and the Balkans. Oelgemöller (2011) illustrates how the term ‘transit’ has emerged in the geo-political space to fulfil a specific political function of suspending unwanted irregular migrants in space. Turkey has been at the forefront of the ‘transit’ project since the early 2000s with the suspending of Iraqi and Iranian migrants in Turkey.

The labelling of countries as ‘transit sites’ often undermines the fact that these countries are also critical destination countries, particularly in the case of Turkey, which is currently hosting the largest number of refugees in the world. Furthermore, the label of ‘transit migrants’ in these countries has a negative perception of clandestine or irregular migrants trying to illegitimately enter Europe and as argued Oelgemöller (2011) demoralises the agency of the migrants. More recently, Greece has also entered into this space in the debate as a ‘transit country’ wherein migrants migrate onwards within the EU from Greece.

Recognizing the limitations of this approach and the loaded constructs that can be associate with the term ‘transit’, in this paper, we use the term ‘transit’ as applied to the countries of Turkey and Greece as spaces of transit migration- which we interpret as a stage in a migrant’s overall journey. We recognize that both Greece and Turkey are simultaneously countries of emigration, immigration and an intended destination choice for some migrants, while simultaneously being spaces of transit for other migrants. In addition, we recognize and seek to highlight in this paper the agency of the respondents in our sample that may in fact not be ‘transit’ migrants in Greece and Turkey.

Migrants decision making factors in transit

The majority of migration research has focused on decision making factors and events in the origin and destination country, but increasingly there is recognition of the migration and refugee journey being in itself an area of investigation (Benezer & Zetter, 2015; Mainwaring & Brigden, 2016). Irregular migration journeys are generally fragmented and broken into different stages (Collyer, 2010), which can become unclear with the journey itself moulding into one event (Mainwaring & Brigden, 2016). It is important to understand the stages of these journeys and the decisions migrants take within their journeys as they often have significant impacts on migrants’ life course. For example, Kaytaz (2016) illustrates that the journeys themselves can be transformational for migrants. In her work, Kaytaz (2016) cites the example of an Afghan migrant who experienced a failed onwards migration attempt and as a result decided to stay in Turkey, which originally was only intended as a country of transit. Specific experiences can therefore change decision making and turn a space of transit into a place of destination.

Limited research has been conducted on how migrants make decisions within this critical stage of the journey. This section provides a brief overview of the factors that can influence these complex decisions. The focus of this paper is on the decision to either stay in the country of transit or to migrate onwards.

Decision making in transit is distinct for three reasons. First, the initial impetus for the migration may have abated in transit. Particularly in reference to forced migrants, this means that if the initial trigger of the migration is no longer of concern individuals will be able to make decisions with more consideration than in a time of crises. Therefore, migrants may choose to stay in the country of transit if they find immediate safety, or they may choose to migrate onwards if they are unsatisfied with the conditions in the transit country. Koser and Pinkerton (2002) highlight that transit countries allow for critical reflection time before making decisions to migrate onwards.

Second, migrants may have access to new sources of information and social networks while in transit that can inform decision making. Collyer (2006) described the new connections made in transit as ‘spontaneous social networks’ meaning individuals encountered during the migration process that are a source of assistance and provide information to migrants on how to survive in transit, in addition to possible travel routes and destinations. At the same time, Suter’s (2012) research has shown the exploitative role of social networks in transit due to the vulnerable conditions of migrants. Networks can thus have positive or adverse consequences for migrants in transit.

Third, capabilities may change in transit. A migrant may only be in a transit country because that is all that they could afford to get to for their migration, aptly stated by Van Hear as ‘I went as far as my money would take me’ (2006). In the transit country, an individual may have more access to resources such as employment opportunities that may lead to gaining enough finances for onwards migration. This reflects on the critical link of capabilities and aspirations in the migration process: ‘People will only migrate if they perceive better opportunities elsewhere and have the capabilities to move’ (De Haas, 2011, 16). Capabilities can affect future planning and if migrants do not have the ability to move onwards they may not aspire to move onwards.

Transit migration adds an additional layer to the ‘destination-origin’ dichotomy that typically exists in migration studies with reference to two countries (Collyer & de Haas, 2012). Migrant decision making in transit encompasses three primary options: to stay in the current country, migrate onwards to another destination, or return to the country of origin (or a previous country of transit or settlement), as compared to the two dimensional dichotomy of stay or go when investigating decision making at origin. An analysis of transit migration should thus encompass multilateral connections and conditions of migrants in the origin, potential destinations, and conditions in the current transit country. That is, decision making in transit brings in an additional stratum to the decision making process of the current conditions in the transit country, as well as the destination and origin country.

Finally, in conceptualizing the decision making process of staying or onwards migration it is important to note that clearly these decisions exist along a continuum. Schapendonk’s (2012a, 2012b) work on migrants in transit showed how settlement can range from tenuously waiting for onwards migration to quite settled lives. The migration decision is also fluid and may be continually changing through dynamic migration trajectories (Wissink et al., 2013).

Greece and Turkey: comparative contexts

The flows of refugees and migrants through Turkey and Greece were unprecedented in 2015 with a record 857,363 arrivals in Greece in 2015 (IOM, 2016a). In 2015 Turkey became the largest global host of refugees (at the time totalling 2.7 million) as Turkey has been steadily receiving Syrian refugees since 2011. There are similarities between these two countries such as both are large countries of destination, transit and sending migrants, but there are several key differences that must be highlighted when considering migrants decision making to stay or migrate onwards from these countries.

Greece and Turkey have substantially different migration policies. In terms of asylum seekers and refugees, Greece is a signatory of the 1951 Geneva Convention on Refugees, and is thereby obliged to accept asylum claims from citizens of all countries in the world. Turkey, on the other hand, is a signatory to the 1951 Geneva Convention on Refugees with a geographical limitation to citizens of European countries.

In 2013, the Turkish Parliament adopted the country’s first migration law, termed the Law on Foreigners and International Protection, which came into force in April 2014 (Kilberg, 2014). This law provides three statuses for humanitarian protection of: refugee (granted only to those coming from Europe); ‘conditional refugee’ (applied to cases of convention refugee outside of Europe); and a status of ‘subsidiary protection’ for those who do not quality as a refugee or conditional refugee but would face death, torture, inhumane treatment or indiscriminate violence upon return. Since coming into force, the status of ‘subsidiary protection’ has been applied to Syrians in Turkey, meaning that Syrians can immediately receive this status in Turkey without going through refugee status determination procedures (Ineli-Ciger, 2015). Other nationalities, including Afghans, Iraqis and Iranians, must apply for ‘conditional refugee’ status in Turkey. Migrants that do receive either subsidiary protection or conditional refugee status cannot necessarily acquire naturalization over the long-term in Turkey, meaning that protection is always temporary (Ineli-Ciger, 2015). With subsidiary protection status, migrants have access to free health care and children can be enrolled in school. They do not have permission to work with this status and need to apply for a labor permission in order to have the right to work.

The implementation of these rights in Turkey has been challenging. At the beginning of 2017, UNICEF reported that 380,000 Syrian children were still missing out on education in Turkey (UNICEF, 2017). Further, Refugees International have highlighted several deficiencies in protection for conditional refugees such as: the requirement to live in satellite cities, access to information and the asylum system, no housing assistance, no access to work permits and legal livelihood opportunities, and insufficient assistance in health care (2017). Some migrants choose to live in Istanbul and travel to the satellite city for signing documentation every two weeks.

UNHCR has a separate procedure in Turkey and can provide access to refugee status determination procedures for resettlement places for the most vulnerable refugees. In 2015, there were 7,577 resettlement departures from Turkey (UNHCR, 2017b), which is far less than one percent of the at concern population in Turkey in 2015. In 2013, UNHCR stopped processing claims from Afghan migrants in Turkey due to capacity issues. UNHCR still registers Afghans, which provides them protection from being deported or detained, but does not give them the option to be considered for resettlement (Dimitriadi, 2015), and thus offers them little protection. UNHCR procedures do not give applicants access to education, nor employment. A key challenge faced by respondents in Turkey was long wait times with UNHCR for interviews.

In Greece, the situation is quite different. The new asylum processing service implemented in 2013 makes claiming asylum easier and faster than under the old process wherein the police administered asylum claims. Asylum applicants receive a ‘pink card’ meaning that they have a form of ID while their claim is processed. Children are also permitted to go to school in Greece if they have refugee or temporary protection status, however, there are no supports provided to asylum seekers or refugees by the government for accommodation, health care, or other living expenses. According to the Greek Refugee Council, in 2014 there were 1,160 accommodation places for asylum seekers in Greece, which compared to the flows was clearly inadequate. This is quite different than in other EU member states, but also highly understandable given the economic crisis in Greece. In 2011, the European Court of Justice issued a ruling that migrants could not be returned to Greece under the Dublin II convention due to the poor conditions in Greece for refugees and asylum seekers. Only recently in 2017 have EU member states started to insist on returning migrants to Greece again under the Dublin convention. Resettlement from UNHCR is evidently not in process in Greece, but as of 2016 the European Commission has agreed to relocate 160,000 asylum seekers from Eritrea, Iraq and Syria who are in Greece and Italy to other EU member states in an effort of responsibility sharing. At the time of data collection this policy was not yet in place and was thus not included in the study as influencing decision making factors.

Finally, it is important to stress that there is a long migration history between Turkey and Greece wherein Turkey has long been subjected to the externalisation of the EU’s migration policies. A culmination of this occurred in March 2016 with the EU-Turkey Agreement to return individuals to Turkey from Greece if they are deemed not to have a legitimate refugee claim under the 1951 Geneva Convention. Although this policy was not in effect at the time of data collection in this study, the EU-Turkey deal has arguably had a significant impact on arrivals in Greece with 162,970 arrivals in Greece for 2016 as of 10 August (IOM, 2016b). At the same time, there are now thousands of migrants stranded in Greece and several questions regarding the moral principles of the deal.

Measuring decision making in transit

In order to assess decision making in transit we analyse four categories of variables: conditions in the country of origin, conditions in the transit country, initial migration destination choice, and the onwards migration journey. First, we operationalize conditions from the country of origin with two variables: country of origin and reason for the initial migration. Research on return migration has demonstrated that reason for the initial migration impacts upon decision making in transit and destination countries in terms of the decision to stay or return (Koser & Kuschminder, 2015). There is a lack of evidence as to how country of origin factors impact upon the decision to migrate onwards or stay in the transit country. We hypothesize that individuals that migrated for security and political reasons are more likely to stay in the current country (H1) as once they have reached a point of safety there is not a necessary need for onwards migration.

Second, we examine conditions in the transit country. There is an increasing recognition that conditions in transit countries can determine if the transit country becomes one of settlement, or stays one of transit. This decision can be based on a variety of factors such as access to employment, education, citizenship, and treatment from locals in the transit country versus the continued perceived risks of onwards migration and life in the intended destination country. A country may have been intended as the destination country, but poor conditions or unrealized ambitions there can subsequently lead to onward migration (Düvell, 2014; Jordan & Düvell, 2002). For example, Brewer and Yükseker (2009) found that in a survey of African migrants in Turkey that discrimination, racism and police violence prevented them from settling in there and prompted onwards migration to Greece. Within the EU, Brekke and Brochmann (2014) found that Eritrean migrants in Italy aspired to migrate onwards to Norway or Sweden because of more generous welfare there. At the same time, Brekke and Brochmann (2014) also found that Eritrean migrants with refugee status were hesitant to attempt migrating onwards due to fear of being returning to Italy under the Dublin II Convention.

Here we look at five variables to measure the current conditions in transit: legal status, subjective living conditions, employment, experiences of abuse and discrimination, and speaking the language. First, we hypothesize that individuals with refugee status or temporary protection are more likely to stay in the current country (H2). Although Dublin II returns to Greece have been suspended by the European Court of Justice, we predict that migrants that have gone through the refugee or temporary protection process and acquired protection status would be more likely to stay in either Greece or Turkey.

Second, we hypothesize thatindividuals that rate their current subjective living conditions as good are more likely to stay (H3). A key driver for onwards migration is poor living conditions and the perception of a better life in the destination country (De Haas, 2011). If migrants therefore perceive their living conditions to be good or very good, we predict that they will be less likely to aspire to migrate onwards.

Third, we hypothesize that individuals that are employed are more likely to migrate onwards (H4). Collyer (2006) finds that if migrants can find work or make network connections to acquire resources, then they are more likely to be able to continue their migration trajectory. If not they may become ‘stuck’ in transit, lacking the capabilities for onwards migration. On the other hand, if a migrant is able to find stable employment they may choose to stay in the country if they can achieve a decent livelihood.

Forth, we hypothesize that individuals that have experienced abuse are more likely to plan to migrate (H5). Brewer and Yükseker (2009) found that in a survey of African migrants in Turkey that discrimination, racism and police violence prevented them from settling in there and prompted onwards migration to Greece. In Greece, high levels of abuse have been reported by migrants from the Golden Dawn extremist right wing political party and from the police.

Fifth, we use the variable ability to speak the local language (Greek or Turkish) as a proxy for integration in the transit country. Speaking the local language infers that the migrant has been in the country long enough to learn the local language. It is often easier to find employment and integrate with locals once one speaks the language and it is also more difficult to envision moving to a new country and learning another language. We hypothesize that individuals that speak the local language are less likely to plan to migrate onwards (H6).

The third category of variables that we examine is the original migration destination choice in order to capture how migrants’ decision making may change in transit. A transit country may in fact be the intended destination and become a transit country or the opposite may occur when a transit country unintentionally becomes the place of ‘settlement’. Onwards migration requires resources and when migrants cannot acquire these resources they may begin to settle, such as Sub-Saharan Africans in Morocco (Collyer, 2006). We operationalize changes in migrant decision making through the variable if the current country of stay was the intended destination in the initial migration. We hypothesize that for individuals where Greece or Turkey (the current country) was their intended destination these individuals are more likely to stay (H7).

Finally, we examine the role of the onwards migration journey in terms of migrant decision making. We use a proxy of failed migration attempt to see if migrants that have attempted to migrate onwards and were not successful are deterred from seeking to migrate onwards. We hypothesize that individuals that have tried to migrate onwards in the past are more likely to want to migrate onwards (H8), meaning that a failed migration attempt does not deter onwards migration.

Data and methods

This paper is based on a dataset of 1056 surveys completed with migrants from Afghanistan, Iran, Iraq, Pakistan and Syria in Athens and Istanbul from April-July 2015. These country of origin groups were selected as they were the top five countries of irregular entry into the European Union at the time. Syrians in Turkey were not able to be interviewed as completing the required approval process (as directed by the Turkish government) was not feasible within the given timeframe for this project. Table 1 shows a breakdown of the sample per country by country of origin.

Table 1. Survey data collection.

	Greece	Turkey	Total
Afghanistan	167	208	375
Iran	59	79	138
Iraq	41	95	136
Pakistan	117	146	263
Syria	144	0	144
Total	528	528	1056

As a large percentage of migrants from these countries of origin are unregistered in each country, simple random sampling was not possible. The sampling strategy was therefore based on first network and intercept point sampling on the street in communities where migrants from each country of origin were known to reside, and second, on snowball sampling from the respondents. The surveys were administered by migrants trained by the research team using a peer approach. The survey was translated into the required native languages prior to the training and conducted using CAPI methods on tablets. After each interview the interviewer asked for the phone number of respondent so that their field supervisor could verify the interview. One third of respondents were called for follow-up by field supervisors to confirm their participation in the survey and to check up to five of their responses. If all responses in the phone check were different from the recorded information by the interviewer the survey was excluded. Further, all interviewers were in touch with the field supervisor and project manager on a regular basis throughout the fieldwork and any challenges were addressed on a regular basis. The use of phone checks and regular management of the fieldworkers allows for confidence in the dataset. The benefits of the migrant to migrant interview approach included that the surveys were conducted directly in native languages without translation, migrants were able to travel inconspicuously to places of interest such as smugglers houses, and migrants were able to create trust for discussing sensitive information.

We also collected 60 follow-up in-depth interviews with participants from the questionnaire (Afghans = 15, Iranians = 12, Iraqis = 9, Pakistanis = 13, Syrians = 11). Individuals were selected for interview based on the criteria of having a mix of employment and unemployment, and a complex case that was identified by the interviewer as having multiple dimensions that were not necessarily all captured well within the survey. These interviews have been used in this paper to assist in verifying and further contextualizing the results.

We conceptualize migrant decision making based on the question At this moment, do you want to: 1) stay in Greece/Turkey, 2) migrate to another country; 3) return to your country of origin; 4) return to the country you were last living in. This question reflects the aspirations of the migrant at the moment, meaning that it is not reflecting their long-term plans, however, nor does the question require that the person currently has concrete plans in place for their onwards migration. Due to the low number of respondents that wanted to return to either their country of origin or country of last settlement (n = 49), we focus in this paper on the decision making between onwards migration and stay and therefore exclude the options of return to your country of origin or return to the country you were last living in from the dependent variable and the analysis.

As the purpose of this paper is to identify factors influencing the decision to migrate onwards, standard probit models were estimated, one for each country. The binary dependent variable takes the value of 1 if the migrant wants to migrate onwards and 0 if the migrant intends to stay in the country of transit. Denoting y_i as the binary outcome variable, Φ as the standard normal distribution function, x_i as the vector of explanatory variables, and β as the vector of coefficients to be estimated, the binary outcome model is given by:

$Pr(y_i=1|x_i)=\theta \left(x_i\beta \right)\mathrm{w}\mathrm{i}\mathrm{t}\mathrm{h}\mathrm{i}=1,\dots ,\mathrm{N}$

In this case, the dependent variable is the probability that an individual will migrate onwards. The models are estimated with robust standard errors and results are presented as average marginal effects. The key independent variables of interest in both models are: country of origin, current migration status, current subjective living conditions, employment status, previous attempts to migrate onwards, if the current country was the initial destination country, if they have experienced abuse, if they migrated with a smuggler to Greece or Turkey, and if they speak the language of the current country of residence. Control variables include sex, marital status, educational status, and whether the migrant comes from an urban or rural area. Further variables were tested in the models, such as duration of stay in current country and age squared, among others; however, due to lack of significance and given that including these variables did not affect the other coefficients, they were not included in the final models.

Descriptive statistics

Tables A1 and A2 in the appendix show the descriptive results of all relevant variables and Table 2 shows descriptive statistics of respondents based on their migration intention (whether they are planning to migrate onwards or to stay). It is striking that the majority of participants in both countries plan to migrate onwards, with more people planning to migrate onwards from Greece (75%) than Turkey (63%). It may be viewed as surprising that respondents reported wanting to migrate onwards from Greece with a 12 percent higher frequency than from Turkey. The analysis section will shed light on this finding in that respondents report poorer subjecting living conditions and lower employment rates in Greece than Turkey, which significantly influence their decision making processes.

Table 2. Descriptive statistics based on decision to migrate onwards or stay in current country.

		Greece			Turkey
		% of the Total	Migrate %	Stay %	% of the Total	Migrate%	Stay%
Country of Origin
Afghanistan		31.6	72.3	27.7	39.3	58.6	41.4
Iran		11.3	75.9	24.1	14.9	42.7	57.3
Iraq		7.8	90.0	10.0	18.2	86.2	13.8
Pakistan		22.1	65.2	34.8	27.6	65.0	35.0
Syria		27.2	82.1	17.9
Reason for initial migration
Security		54.9	80.5	19.5	38.1	62.3	37.7
Risk/no right to work		18.7	69.1	30.9	8.0	78.1	21.9
Lack of employment /education		15.6	65.9	34.2	33.8	58.9	41.1
Family reasons		5.7	75.9	24.1	13.7	66.2	33.9
Other		5.1	72.0	28.0	6.5	60.6	39.4
Current migration status
Refugee		18.2	60.6	39.4	4.7	58.3	41.7
Asylum seeker		25.5	64.1	35.9	23.1	64.5	35.5
Irregular		28.5	77.3	22.7	62.6	63.8	36.2
Temporary protection		25.1	94.6	5.4
Other		2.7	71.4	28.6	9.6	57.5	42.6
Subjective current living conditions
		41.8	55.3	44.8	54.4	57.3	42.8
Currently employed
	31.8		52.4	47.6	54.6	51.4	48.6
Language
	41.4		52.3	47.7	33.1	55.2	44.8
Experienced abuse or discrimination
	62.2		73.5	26.5	32.1	68.9	31.1
If the current country is the intended destination country of stay
	18.7		37.9	62.1	47.6	39.2	60.8
Have attempted to migrate onwards
		39.5	78.0	22.0	12.5	81.0	19.0
Migrated with a smuggler
		76.1	77.0	23.0	63.1	62.4	37.6
Married		32.7	77.2	22.8	31.2	56.8	43.2
Education
No formal education		12.9	70.2	29.8	11.9	48.2	51.8
Primary		26.3	76.1	23.9	28.0	63.0	37.0
Secondary		29.5	77.3	22.7	32.5	72.4	27.6
Vocational/higher		31.4	74.4	25.6	27.6	58.6	41.4
Urban		64.5	75.2	24.8	70.1	65.6	34.4
Total		100	75.1	24.9	100	63.1	36.9

Source: authors’ calculations.

Regarding the first set of independent variables relating to conditions in the country of origin, Iraqis are the most likely to plan to migrate onwards in both countries (90% in Greece and 86% in Turkey) (cautioning however that the sample of Iraqis interviewed was relative small), followed by Syrians (82%) and Iranians (76%) in Greece and by Pakistanis (65%) and Afghans (59%) in Turkey. Turning to the reason for initial migration (security, no right to work, lack of employment or educational opportunities, family reasons, or other reasons), a higher percentage of participants in each category are planning to migrate onwards than to stay. In two cases the difference is quite notable, such as the case of those who migrated for security and family reasons in Greece (80% migrate onwards versus 20% stay) and for risk of deportation, police harassment, or no right to work in Turkey (78% migrate onwards versus 22% stay).

The second set of independent variables we examine relates to conditions in the current country of transit. Migration status is self-reported by the respondents and has been categorized as refugee, asylum seeker, irregular, temporary protection, and other status, which could include student or other visa holder. Although in both countries temporary protection is possible, no respondents stated having temporary protection in Turkey. A higher percentage of migrants, regardless of their current status in the country, intend to migrate onwards. It is surprising that 94.6 percent of temporary protection holders seek to migrate onwards from Greece.

Respondents were asked to assess their current subjective living conditions based on a five-point scale of bad, very bad, average, good, or very good. Based on this variable we created a dummy variable where 0 represents bad or very bad living conditions and 1 represents average, good or very good conditions. Interestingly, a higher percentage of participants in Turkey (54%) reported having average or good living conditions as compared to less than 42 percent in Greece. Among those that perceive their current living conditions as average or good, a similar percentage of individuals plan to migrate onwards than to stay in both countries (55% in Greece and 69% in Turkey).

A higher percentage of participants were employed in Turkey (55%) than in Greece (32%), however, nearly all participants employed in Turkey were working irregularly (97%), whereas one third of participants that were employed in Greece were working legally (11%). There is little variation between employed participants in terms of wanting to migrate onwards or stay in both countries. Amongst those that were unemployed, the majority plan to migrate onwards versus stay in both countries. A slightly higher percentage of participants that speak Greek reported their intention to migrate onwards from Greece, and the same occurs in Turkey.

It is disconcerting that 62 percent of participants in Greece and 52 percent in Turkey reported experiencing physical or verbal abuse since their arrival in the respective country. A high percentage of those experiencing abuse or discrimination plan to migrate onwards from both countries (74% in Greece and 69% in Turkey).

Next we examine if the initial migration destination choice changed or not in transit. For almost half of the participants in Turkey, Turkey was their intended destination country when they left their country of last residence. This number was much lower in Greece with nearly 19 percent of participants intending to migrate to Greece. Half of the Pakistanis and 19 percent of the Syrians interviewed in Greece were intending to migrate to this country, but this was less than ten percent for all other groups. In Turkey, Iranians (63%) and Pakistanis (60%) were most likely to seek to migrate to Turkey, followed by Afghans (38%) and Iraqis (33%). In both countries, slightly over 60% of the participants that stated Greece or Turkey as their intended destination country when they left their country were planning to stay.

The final independent variable examines if respondents have had a previous attempt to migrate onwards. A higher percentage of participants had attempted to migrate onwards from Greece (39.5%) than from Turkey (12.5%). It is important to note again that the fieldwork in Greece was conducted in May-June 2015, prior to the opening of the northern border with the Republic of Macedonia. At the time of interview, and in the years prior to this when some of the respondents had made these attempts, migrating onwards from Greece was difficult. In the qualitative interviews, respondents reported being pushed back to Greece from Macedonia by border police, being arrested at the airport, and being arrested while trying to get in shipping containers to reach Italy. The work of Dimitriadi (2015) further details the difficulties of Afghan migrants in trying to migrate onwards from Greece. In Greece participants also had a slightly higher number of migration attempts with an average of 4.4 compared to an average of 2.3 migration attempts from Turkey. Approximately 80 percent of those who had attempted to migrate onwards are more likely to continue to migrate onwards in both countries.

Turning to the control variables, a larger number of participants that were married are planning to migrate onwards than to stay, albeit to a greater extent in Greece (77%) than in Turkey (57%). Planning to migrate onwards does not appear to vary greatly in Greece based on the level of education. In Turkey, a higher percentage of participants with secondary education intend to migrate onwards than to stay, whereas among participants with lower levels of education or participants with higher or vocational education differences between the number of individuals wanting to migrate onwards or to stay are not so high. Finally, in both countries a larger number of participants from an urban background want to migrate onwards.

The number of women interviewed in both countries was quite small at only eight percent of the sample (41 in Greece and 39 in Turkey). These migration flows do tend to be gendered, however, not necessarily to this extent. Eurostat data shows that the percentage of female asylum seekers in the EU-28 in 2015 for each country of origin group was 19 percent for Afghans, 25 percent for Iraqis, 26 percent for Iranians, five percent for Pakistanis and 29 percent for Syrians (Eurostat, 2016). These figures do not necessarily represent irregular migration flows that this study intends to capture, but can be considered as an indication of the gendered dimension of these migration flows. As respondents were approached in the street using intercept point sampling, more men were available for interviews than women. Efforts were made to snowball to find further female respondents, but these attempts were only somewhat successful. The result is that a key limitation of the study results is the highly male sample and a gendered analysis cannot be conducted.

Results

Table 3 presents the regression results and shows how different factors correlate with the intention to migrate onwards. Regarding country of origin, one can see that as compared to Afghans, in Greece all other country of origin groups are slightly more likely to migrate onwards by 11 to 17 percentage points. In Turkey, Iraqis and Pakistanis are 39 and 26 percentage points, respectively, more likely to migrate onwards as compared to Afghans. On the other hand, Iranians and Afghans do not significantly differ in their intention to migrate onwards from Turkey.

Table 3. Factors influencing the decision to migrate onwards or stay and settle.

Greece		Turkey
Country of origin (ref category: Afghanistan)
Iran	0.16**	0.03
	(0.06)	(0.08)
Iraq	0.15^+	0.39**
	(0.06)	(0.07)
Pakistan	0.11*	0.27**
	(0.05)	(0.05)
Syria	0.11*
	(0.05)
Reason for migration (ref category: security)
Risk/no right to work or stay	−0.02	0.14*
	(0.04)	(0.08)
Family reasons	0.02	−0.02
	(0.06)	(0.06)
Other reasons	−0.07	0.02
	(0.07)	(0.08)
Lack of employment/educational opportunities	−0.02	0.04
	(0.04)	(0.05)
Status (ref category: Asylum seeker)
Refugee	0.04	−0.00
	(0.04)	(0.09)
Irregular	0.06	0.09
	(0.04)	(0.05)
Temporary protection	0.04
	(0.06)
Other	0.09	−0.06
	(0.09)	(0.08)
Current living conditions	−0.20**	−0.09*
	(0.03)	(0.04)
Employed	−0.03	−0.11*
	(0.04)	(0.04)
Speaks Greek/Turkish	−0.19**	0.02
	(0.04)	(0.04)
Abused	−0.07*	0.04
	(0.05)	(0.05)
Attempted to migrate onwards	0.00	0.13*
	(0.04)	(0.06)
Greece/Turkey was the intended destination	−0.24**	−0.31**
	(0.04)	(0.03)
Migrated with a smuggler	0.06^+	0.06
	(0.04)	(0.07)
Married	−0.02	−0.06
	(0.04)	(0.05)
Male	−0.06	−0.07
	(0.07)	(0.07)
Education (ref category: no education)
Primary	0.03	0.04
	(0.05)	(0.06)
Secondary	0.06	0.11^+
	(0.05)	(0.06)
Vocational/higher	−0.01	0.05
	(0.05)	(0.07)
Urban	−0.02	0.12*
	(0.03)	(0.05)
Observations	514	487
Pseudo R^2	0.39	0.29

Source: authors calculations.

The second variable representing conditions in the country of origin, reason for the initial migration, is not significant in Greece, but is somewhat significant in Turkey. In the latter, participants that stated leaving the country where they were last living for the reasons of either risk of deportation, no right to study or work, or police harassment are more likely to seek to seek to migrate onwards. The participants that selected these reasons for their migration are primarily Afghans that have previously been living in Iran (57%).

Turning to the conditions in the current country of stay, several variables were significant; however, in both Greece and Turkey, migration status is not significantly correlated with the decision to migrate onwards. This is surprising, as it can be assumed that migrants that receive a regularization status such as refugee or temporary protection are less likely to desire to migrate onwards. However, in the Greek context receiving refugee or temporary protection status does not entitle migrants and refugees to any form of social protection, meaning that they are still required to provide for themselves and their families.

Robust standard errors in parentheses; +p-value<10%, * p-value<5%; ** p-value<1%

Participants’ current subjective living conditions are significant in both country contexts. In Greece, participants that consider their living conditions as good or average are 20 percentage points less likely to seek to migrate onwards. In Turkey, participants that consider their living conditions as good or average are nine percentage points less likely to seek to migrate onwards.

A second variable examining migrants’ conditions in the current country was if the participant was currently employed. In Greece, this variable is insignificant, but in Turkey, employed participants are eleven percentage points less likely to migrate onwards as compared to participants that want to stay in the country. In Greece, participants that speak Greek are less likely to want to migrate onwards, whereas speaking Turkish is insignificant in the Turkey model. Further, participants that have experienced at least one form of abuse in Greece are less likely to seek to migrate onwards. This will be discussed further in the discussion.

Having previously attempted to migrate onwards is not significant in the Greek model. In Turkey, however, participants that have previously attempted to migrate onwards are twelve percentage points more likely to want to migrate onwards. In both countries, participants whose intended destination has been either Greece or Turkey are significantly less likely to plan to migrate onwards by 23 and 30 percentage points respectively. Finally, in Greece, participants that had migrated with a smuggler to Greece were seven percentage points more likely to seek to migrate onwards, whereas this variable was insignificant in Turkey.

In Greece, none of the controls of married, age, education, male or urban are significant. In Turkey, participants with secondary education are slightly more likely to migrate onwards than to stay, and participants from an urban background are also more likely to plan to migrate onwards.

Discussion

The results demonstrate that several factors are important in respondents’ decision making factors for onwards migration from transit countries. First, we hypothesized that individuals that migrated for security and political reasons are more likely to stay in the current country (H1) based on the idea that they had reached a place of safety. The results show that this is not the case and the hypothesis needs to be discarded. This can perhaps be explained by other variables in the analysis such as the conditions in the current country of stay.

In the current country of stay, we find that current subjective living conditions are highly significant in the decision making in both Greece and Turkey. We can therefore accept H3 that individuals that rate their current subjective wellbeing as good are more likely to stay (H3). This supports the idea that onwards migration is driven by poor living conditions and the perception of a better life in the destination country (De Haas, 2011).

In the transit country, we next examine the role of legal status in the migration decision and we hypothesize that individuals with refugee status or temporary protection are more likely to stay in the current country (H2). The results demonstrate that legal status is insignificant in influencing the migration decision. Potential reasons for this include that first, having legal statuses such as refugee or temporary protection status does not enable migrants to work in Turkey, or to any forms of social protection in Greece. We would have expected that subjective living conditions and legal status were highly correlated, but in this case they are not. Therefore, the expected benefits of being legalized are not necessarily realized in these contexts.

Employment also proved to be highly significant in the case of Turkey, but not in the case of Greece. In Turkey, individuals that were employed were less likely to seek to migrate onwards, which contradicts hypothesis statement four. Even though more than 95% of migrants working in Turkey were employed irregularly, they were still more likely to plan to stay. This demonstrates that employment in Turkey is a critical component of attracting people to stay. Employment is essential for migrants to be able to provide for themselves and their families’ livelihoods in Turkey and without employment migrants are highly vulnerable. We also find that migrants that speak Greek, which is a proxy for length of time spent in the current country, are significantly more likely to seek to stay in Greece, which partially confirms hypothesis statement six.

We expected that individuals that have experienced abuse are more likely to plan to migrate (H5); however, the results in Greece presented the opposite that migrants that had experienced abuse were more likely to seek to stay in Greece. One potential explanation for this comes from the psychological literature of the concept ‘learned helplessness’ or feelings of ‘entrapment’ leading to giving up on migration aspirations (Bhugra & Ayonrinde, 2004; Gilbert & Allan, 1998). According to Bhugra and Ayonrinde (2004) migrants that experience arrested flight leads to the submission of explorative behaviours, meaning that migrants are more willing to accept their current environment including the stresses (in this case the experiences of abuse), versus trying to continue their migration aspiration.

Turning to the original migration destination choice, we confirm hypothesis seven that in both countries for individuals where Greece or Turkey (the current country) was their intended destination these individuals are more likely to stay (H7). This suggests that in contrast to some of the qualitative literature on migrants’ decision making in transit that decisions regarding onwards migration may not be frequently changing. It is recognized, however, that a limitation of this study is that the data only represents one moment in time, and further longitudinal research would be required to make this finding more robust.

Finally, we find that a failed migration attempt does not deter onwards migration from Turkey and we can partially confirm hypothesis statement eight. The main reasons cited by migrants for a failed migration attempt in Turkey were being apprehended by the police (39%) and being returned to Turkey by authorities in another country (35%). This illustrates that migration controls and active policies to deter migration, do not thwart migration aspirations in Turkey.

Conclusion

This paper has used a unique dataset to assess migrants’ decision making processes to migrate onwards or stay in Greece and Turkey in the context of the beginning of the mass influx of flows to Greece occurring in late spring and early summer 2015. It goes beyond looking at the socio-legal context for decision making, and incorporates four different categories of influencing variables: conditions in the country of origin, conditions in the transit country, initial migration destination choice, and the onwards migration journey. For migrants originating from Asia, both Greece and Turkey are essential crossing points in their journeys to Europe, wherein stay in these countries can range from a few days to years.

The results of this study illustrate that the majority of respondents surveyed in Greece and Turkey (that had been living in each country for a range of a few days to several years) seek to migrate onwards from these countries. On the one hand, this is not necessarily surprising given the unprecedented irregular migration inflows to Europe in the second half of 2015 and early 2016. On the other hand, it is quite striking and concerning the sheer volume that prefer onwards migration to stay, and in particular that a higher number of migrants want to migrate onwards from Greece than Turkey. Three central reasons for this found in the study were: first, respondents in Greece were more likely to rate their subjective living conditions as poor; second, migrants in Greece were less likely to be employed than respondents in Turkey; and third, Turkey was considerably more frequently the intended destination in the migration episode than Greece. All of these factors were highly significant in influencing the migration decision and illustrate that poor living conditions, lack of employment opportunities, and initial migration destination intentions are central reasons that drive the desire to move onwards from Greece. Therefore, the influencing variable that was most significant in the decision making was conditions in the current transit country. This suggests an important finding for future research that stresses the salience of conditions in countries of stay on migration journeys. Understanding these contexts in more detail and how to improve conditions for migrants in an important area of further investigation.

A central finding of this study builds on the already existing evidence as to the concerns over conditions for asylum seekers and migrants in Greece. At the time of the fieldwork in this study it was difficult to migrate beyond Greece, which changed in August 2015 with the opening of the northern border. In March 2016, that border was again closed and a reported 57,000 migrants were trapped in Greece in August 2016. In 2019, there were still 15,000 migrants living in camps on the Greek islands. These individuals may also want to migrate onwards, but are stuck in a system that does not have the capacity to provide for them as the Greek authorities and the EU continue to fall short on their abilities to provide housing to the migrants and transfer people to the mainland. The current economic crisis in Greece does not enable the country to be able to provide for social protection including housing and the daily needs of this group. As shown in the results, even respondents with refugee and temporary protection status in Greece wanted to migrate onwards due to the poor conditions and lack of supports for refugees in the country. Although Greece is an EU country and upholds the values of the Convention and rights of refugees, it is not able to create the necessary conditions for refugees to stay. This finding reiterates the need for further burden sharing in the EU and support to Greece for being at the frontline of asylum seekers and migration to the EU. The EU relocation scheme was slow to start, but in 2017 gained momentum with higher number of migrants being relocated, however, the end result was far from enough to make a meaningful difference in Greece’s caseload.

In Turkey, with the implementation of the EU-Turkey deal in 2016, migrants have essentially become locked in place, meaning that those seeking to migrate onwards from Turkey now have very limited options to do so. This is also an area of concern given the conditions reported in Turkey in this study and others. The results here stress the need for further access to employment in Turkey, as those that are employed were significantly less likely to seek to migrate onwards. Employment is vital in being able to have a decent life and survival in Turkey. In 2019, 19,000 Syrians were reported to have received work permits in Turkey. Although substantial progress, this is also far less than the need. In February 2016 at the Syrian Donors Conference in London David Miliband proposed 600,000 work permits for Syrian refugees in Turkey (Wintour & Black, 2016). Scholars in Turkey have been calling for migrants’ access to the labour market as a critical component of integration for several years (Icduygu, 2015). By providing access to employment, migrants may choose to stay in countries such as Turkey. This is also a critical component in framing our language to factors that enable migrants to stay, versus preventing them from onwards migration.

Finally, the results highlight that decision making in transit differs from the original migration decision and needs to be examined as a different process. The conditions of the transit country add in an additional layer to the migration decision and the results of this paper have demonstrated that the conditions in transit are highly significant in migrants’ decision making processes. Further research is required to understand migrants’ decision making across the origin-transit-destination countries and in particular at this complex stage, or transit stages, of the migration journey.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This research was conducted as part of the Irregular Migrants Decision Making Factors in Transit Project funded by the Australian Department of Immigration and Border Protection and the Australian National University’s Collaborative Research Programme on the International Movement of People. The views expressed in this paper are those of the authors and do not represent those of the Department of Immigration and Border Protection or the Australian National University.

Notes on contributors

Katie Kuschminder

Katie Kuschminder is an Assistant Professor at the Maastricht Graduate School of Governance and United Nations University Merit.

Jennifer Waidler

Jennifer Waidler is a Research Analyst at the UNICEF Office of Research – Innocenti, where she evaluates cash transfers and cash plus programs in Sub-Saharan Africa. She holds a PhD in Economics and Governance from UNU-MERIT / Maastricht University and a Master’s in Public Policy with a specialization in Social Policy Design from the same university. Her main areas of interest/ expertise include multidimensional poverty measurement, effects of migration and remittances, and impacts of cash transfer programs on outcomes such as food security, subjective well-being, and expenditure behaviour.

Appendix 1

Table A1. Descriptive statistics of relevant variables (Greece).

Variable	N	Mean	SD	Min	Max
Migrate onwards	519	0.75	0.43	0	1
Country of origin
Afghanistan	529	0.13	0.33	0	1
Iran	529	0.11	0.32	0	1
Iraq	529	0.08	0.27	0	1
Pakistan	529	0.22	0.42	0	1
Syria	529	0.27	0.45	0	1
Current migration status
Refugee	529	0.18	0.39	0	1
Asylum seeker	529	0.26	0.44	0	1
Irregular	529	0.29	0.45	0	1
Temporary protection	529	0.25	0.43	0	1
Other	529	0.03	0.16	0	1
Subjective current living conditions	529	0.42	0.49	0	1
Currently employed	529	0.32	0.47	0	1
The current country is the intended destination country of stay	529	0.19	0.40	0	1
Have attempted to migrate onwards	529	0.40	0.49	0	1
Experienced abuse or discrimination	529	0.62	0.49	0	1
Language	529	0.41	0.49	0	1
Migrated with a smuggler	528	0.76	0.43	0	1
Reason for initial migration
Security	525	0.55	0.50	0	1
Risk/no right to work	525	0.19	0.39	0	1
Lack of employment /education	525	0.16	0.36	0	1
Family reasons	525	0.06	0.23	0	1
Other	525	0.05	0.22	0	1
Married	529	0.33	0.47	0	1
Education
No formal education	529	0.13	0.47	0	1
Primary	529	0.26	0.44	0	1
Secondary	529	0.29	0.46	0	1
Vocational/higher	529	0.31	0.46	0	1
Urban	529	0.64	0.48	0	1
Male	529	0.92	0.27	0	1

Appendix 2

Table A2. Descriptive statistics of relevant variables (Turkey).

Variable	N	Mean	SD	Min	Max
Migrate onwards	490	0.63	0.48	0	1
Country of origin
Afghanistan	529	0.39	0.49	0	1
Iran	529	0.15	0.36	0	1
Iraq	529	0.18	0.39	0	1
Pakistan	529	0.28	0.45	0	1
Current migration status
Refugee	529	0.05	0.21	0	1
Asylum seeker	529	0.23	0.42	0	1
Irregular	529	0.63	0.48	0	1
Subjective current living conditions	529	0.54	0.50	0	1
Currently employed	529	0.55	0.50	0	1
The current country is the intended destination country of stay	529	0.48	0.50	0	1
Have attempted to migrate onwards	529	0.13	0.33	0	1
Experienced abuse or discrimination	529	0.32	0.47	0	1
Language	529	0.33	0.47	0	1
Migrated with a smuggler	529	0.63	0.48	0	1
Reason for initial migration
Security	527	0.38	0.49	0	1
Risk/no right to work	527	0.12	0.33	0	1
Lack of employment /education	527	0.34	0.47	0	1
Family reasons	527	0.10	0.30	0	1
Other	527	0.06	0.24	0	1
Married	529	0.31	0.46	0	1
Education
No formal education	529	0.12	0.32	0	1
Primary	529	0.28	0.45	0	1
Secondary	529	0.33	0.47	0	1
Vocational/higher	529	0.28	0.45	0	1
Urban	529	0.70	0.46	0	1
Male	528	0.93	0.26	0	1