Evaluations of Return Within A Mass Deportation: Ethiopians’ Experiences of Return After Expulsion From Saudi Arabia

Abstract

This article seeks to understand the role of the migration lifecycle in the subjective evaluation of return by Ethiopian deportees from Saudi Arabia, focusing on the conditions that lead to positive evaluation of the return. Logistic regression analysis was carried out on a unique data set of 2,039 Ethiopian deportees from Saudi Arabia collected in 2014. Despite having been deported, 45 per cent of respondents evaluate their return positively. It is found that subjective socio-economic position is highly influential in returnees’ assessment of their reintegration, but those who are self-employed on return are much less likely than those who are unemployed to describe their return positively. This may be linked to work conditions and because these respondents are engaged in necessity entrepreneurship. Additionally, all stages of the migration cycle, including before and during the migration episode, are influential in shaping perceptions of return.

INTRODUCTION

In the context of deportation, the term “reintegration” has been challenged as it has been illustrated that deportation is not necessarily returning “home” to a place of safety and security (Khosravi, 2018). Following from the introduction of this special issue, it is recognized that reintegration is an adjustment process that occurs over time and, in fact, may not occur for some returnees or deportees. This article focuses on the lesser researched area of subjective reintegration by examining returnees’, and primarily deportees’, perceptions of return. Reintegration can be studied from either an objective position – which focuses on certain economic conditions, or a subjective position – meaning the perceptions of the returnee (Gmelch, 1980). We focus on the subjective perceptions of returnees, which has received less attention in the literature. The few studies that have focused on perceptions of return (Gmelch and Gmelch, 1995; Ammassari, 2009 ; Wong, 2014; Ruiz, Siegel and Vargas-Silva, 2015) generally do so with voluntary returnees or refugee repatriates. These studies have found that subjective perceptions of return may differ from objective reintegration measures (such as employment status), stressing the importance of research on subjective perceptions of return. None of these studies have addressed deportees’ perceptions of their return. This article makes an original contribution by intersecting both reintegration research and deportation studies through a case study of Ethiopian deportees from Saudi Arabia.

There is an increasing body of research that examines returnees’ perceptions of return (Fransen et al., 2017; Bilgili et al., 2018 ; Fransen and Bilgili, 2018). Within this research, several authors acknowledge the importance of the migration life cycle in shaping return perceptions and experiences (Cassarino, 2014; Bilgili et al., 2018). For example, Cassarino demonstrates through his preparedness model that migrants’ patterns of reintegration are impacted by the conditions in the country of origin/return, the migration experiences, and the conditions of the return episode (Cassarino, 2014). Bilgili et al. (2018) show that returnees’ reintegration strategies are determined both by their migration cycle and experiences abroad, which impact their cultural orientation and self-identification upon return, thus influencing their reintegration and perceptions of return. Deportees experience return as a life cycle interruption, meaning that they are unable to prepare for their return and have difficult reintegration experiences (Cassarino, 2014; Khosravi, 2018). There is rich anthropological research on deportees that highlights the negative experiences of feelings of abandonment and the challenges faced post-return (Khosravi, 2018). Issues of shame and vulnerability are frequently demonstrated in this literature (Schuster and Majidi, 2013; Golash-Boza, 2015).

Drawing on reintegration research and deportation studies, this article makes a unique contribution to the field through a quantitative analysis of a data set of primarily forced returnees to Ethiopia from Saudi Arabia. In late 2013 to early 2014, over 163,000 people were forcibly removed from Saudi Arabia to Ethiopia. The deportations were quick and resulted in a mass return to Ethiopia that became a humanitarian emergency. The Government of Ethiopia and international organizations were swiftly engaged to respond to the mass deportations to support returnees in returning to their communities and reintegrating. This paper relies on a data set of 2,039 Ethiopian returnees who were primarily deported from Saudi Arabia. The data were collected in June 2014 for the International Labour Organization (ILO). The data and case analysis in this paper provide a unique contribution to examining a case of mass deportation with a large data set of returnees.

In this paper, we examine factors that influence returnees’ short-term reintegration experiences based on their subjective evaluation of return. We aim to understand the role of the migration lifecycle in returnees’ subjective evaluation of their return and the conditions that lead to positive evaluation of the return. We consider short-term reintegration as being up to eight months post-return, and subjective evaluation is measured by returnees’ perceptions of their return experience. Following from the literature, we examine four groups of variables related to the migration life cycle; life prior to migration, the migration experience, return conditions and support received post-return. Through this analysis, we aim to increase understandings of the factors that influence short-term reintegration in a situation of mass deportations.

THE MIGRATION LIFE CYCLE

The migration life cycle is highly relevant to the process of forced return and reintegration. In the first instance, systems of deportation place emphasis on the returnees’ previous social and economic ties in the country where they are deported. The vocabulary of “reintegration” assumes that the subject was integrated before their migration episode took place, imposing notions of “home” that in some cases will be inappropriate (Khosravi, 2018). Nevertheless, Kuschminder (2016) found in a study of assisted voluntary returnees that those who returned to their communities of origin were more likely to have sustainable return. Conditions before the migration also have a knock-on effect on reintegration through the migrants’ experiences abroad, as better access to funds and social networks may help migrants to have better livelihoods abroad. Thus, we hypothesize that returnees who had a better socio-economic position prior to migration are more likely to have a positive evaluation of their return (H1).

Second, the experience abroad has been demonstrated to have a significant effect on the return experience (Kuschminder 2017). Neoclassical and New Economics of Labour Migration (NELM) theories describe that positive experiences abroad, specifically higher wages or stable employment, define the perceived success of a migration episode (Massey et al., 2009). Therefore, in cases of deportation the migrant may be considered a “failed migrant” if they are returned before fulfilling their economic goals. Indeed, the ability to utilize savings from work abroad may be crucial to the returnees’ ability to secure a livelihood (Nisrane et al., 2017). We hypothesize that respondents with a good or very good socio-economic position during migration will have a more negative experience of return as their cycle and earning potential was interrupted (H2).

In addition to the economic resources, migrant workers will have had a variety of social and cultural experiences while abroad and may have had the opportunity to acquire professional skills. This is often the case regardless of legal status. The fact that the migrant may learn skills and cultural mores appropriate to the deporting country has led to an idea of “deportation as development.” However, as Collyer (2018) emphasizes, the positive effects of the migration are dependent on the wider economic context and may even be reversed in cases of forced return. On the other hand, a migration episode could have long-lasting negative effects on migrants’ physical and psychosocial well-being (Turnbull, 2018). Migrant workers are frequently regarded as “disposable labour” and their human rights not respected. Indeed, the process of deportation itself is dehumanizing, including incarceration and other punitive measures. We hypothesize that respondents who have faced a challenge abroad will have a more negative evaluation of their return (H3) and that respondents who have gained a professional skill abroad will have a more negative evaluation of their return (H4). This follows from the previous hypothesis in that we assume that people who were doing well abroad will have a more difficult return experience.

Third, the conditions of return are important in influencing the return experience. As highlighted by Ruben et al. (2009), the process of becoming embedded in a society is determined by the interaction of many conditions upon return, including economic, social and psychosocial dimensions. In cases of forced return, migrants mostly do not have the opportunity to bring back accumulated wealth, which could be detrimental to the positive effects that a migration episode can have (Cassarino, 2014; Collyer, 2018). Deportation can have negative connotations, with returned migrants treated as virtual foreigners or sometimes with suspicions of criminality (Schuster and Majidi, 2013). These factors are added layers of challenge in establishing a livelihood to those who already experience discrimination on the grounds of characteristics such as ethnicity, religion or gender (Khosravi, 2016; De Regt and Tafesse, 2016; Bilgili et al., 2018). We hypothesize that respondents with employment (H5) and/ or support from family, friends and their community, or are a member of an association (H6) will have a more positive evaluation of their return.

Finally, post-return assistance plays a significant part in the management of deportation, both completing the institutional process and revealing how deportation is conceptualized by governments, NGOs and development agencies. On the one hand, post-return assistance is practical, often targeting additionally vulnerable groups by providing, for example, women-only spaces or vocational training. On the other hand, services impose ideas of migrants as entrepreneurs, and “home-coming” through the common provision of entrepreneurship training programmes and transport assistance, sometimes forcefully reuniting deportees with their families (Ahumuza Onyoin, 2017; Schuster & Majidi, 2013). The emphasis on reunification and entrepreneurship schemes can be seen as depoliticizing deportation; by focusing on the return in a positive light, it obscures the brutality of the process (Lecadet and Tafesse Melkamu, 2016; Khosravi, 2018, p. 11). Furthermore, Khosravi (2018) argues that deportees’ lives upon return are characterized by “waiting”, and often a high dependency on international organizations and NGOs emerges as deportees wait for support. In this case, we recognize that support upon return may either positively or negatively reflect on the return experience and thus leave this open to exploration in the model (H7).

THE CONTEXT OF RETURN AND DEPORTATION FROM SAUDI ARABIA TO ETHIOPIA

There have been large increases in migration from Ethiopia to the Middle East over the last decade. The Ethiopian Ministry of Labour and Social Affairs (MoLSA) estimate that the number of documented migrants who departed for Arab countries increased from 30,000 to 200,000, from 2009 to 2012 alone (cited in Zewdu, 2018, p.11). However, MoLSA stated these figures may only have represented 30–40 per cent of those who travelled to the Middle East, with the rest migrating through unofficial channels (Frouws, 2014, p. 4). In a move intended to safeguard the well-being of citizens, in October 2013 a ban was put in place that forbade Ethiopians from travelling abroad in search of low-skilled work, but migration continued through irregular means. The ban was lifted in January 2018 following new regulations for employment agencies and the implementation of bilateral agreements with key destination countries, including Saudi Arabia. Nevertheless, the numbers of Ethiopians travelling to the Middle East irregularly may be increasing. IOM estimates that 159,838 Ethiopians and Somalis travelled irregularly through Yemen in 2018, a 60 per cent increase on the total number in 2017 (Mixed Migration Centre, 2019).

There are numerous drivers of migration in Ethiopia. Even though Ethiopia has experienced high levels of economic growth, with GDP growth an average of 10.5 per cent between 2003 and 2016, it remains among the world’s poorest countries (World Bank, 2020). In 2018, Ethiopia was ranked 163 out of 189 countries by the Human Development Index (UNDP, 2018b), and there were 22 million people still living under the poverty line (UNDP, 2018a). As a result of rapid population growth, there are over one million joining the workforce each year. As wages are much higher abroad, many Ethiopians regard migration as the only opportunity to achieve better living standards.

Problems of unemployment in Ethiopia effect women disproportionately; in 2013, the unemployment rate as a percentage of the work force was almost double for women as for men (World Bank, 2020). In rural areas, migration is one of the few options available to women after completing their education. Many choose to migrate in order to gain more independence and as a way to delay early marriage (Schewel, 2018a, 2018b). Surveys conducted in 2011 found that between 54 and 60 per cent of flows to the Middle East from Ethiopia were female (Kuschminder and Siegel, 2014).

Men and women broadly have different experiences of migration to the Middle East. While men often work as guards, daily labourers and on farms, women almost exclusively undertake domestic work (de Regt and Tafesse, 2016). Domestic workers are vulnerable to a range of abuses in private settings; including overwork, forced confinement, non-payment of wages, food deprivation, and psychological, physical and sexual abuse (Human Rights Watch, 2019, p. 497). As male migrants are less likely to have had the opportunity to migrate through legal means, a higher proportion of returnees and deportees have been male (Admassie et al., 2017; IOM, 2019; Mixed Migration Centre, 2019).

In November 2013, Saudi Arabia began to implement strong measures to engineer the “Saudization” of the labour market (Frouws, 2014, p. 53). Following a deadline for amnesty, the authorities began to carry out labour inspections and arrest migrants. The Government of Ethiopia initially expected 30,000 deportees, but the numbers quickly soared to 8,000 arrivals a day, and by December 2013, there had been 100,000 arrivals at Bole International Airport in Addis Ababa (IOM, 2014, p. 12). In total, 163,018 Ethiopians were deported from Saudi Arabia between November 2013 and March 2014 (Mixed Migration Centre, 2019). There was violence against migrants by authorities and private citizens, and inadequate detention facilities (Human Rights Watch, 2015). De Regt and Tafesse (2016) found that a third of returnees had belongings confiscated or were not given the opportunity to bring home their belongings.

The large numbers of deportees prompted a humanitarian response. A “mass migration” movement differs from individual migration by the sheer number of people involved in the movement (Pok, 2012). Mass deportations are most commonly discussed in the literature in the case of the United States (Golash-Boza, 2015). According to data from the Global Detention project, the United States expelled 446,223 people in 2016 (Global Detention Project, 2019b). Saudi Arabia by comparison expelled 525,871 people in 2018, illustrating that Saudi Arabia is also a significant global deporting state (Global Detention Project, 2019a).

On arrival in Ethiopia, the deportees were assisted by the Government of Ethiopia and members of the Ethiopia Red Cross Society, who briefed the migrants on their return and supplied them with pink cards that identified them as eligible for support. The deportees were then led to IOM registration and processing centres where they were given water and food, and a transportation cash allowance to enable them to travel home. As highlighted by Lecadet and Tafesse Melkamu (2016), this particular mass expulsion was seen as a test case for reintegration programmes following a mass deportation.

Following the immediate period of arrivals, support for the reintegration of returnees was coordinated by the ILO in partnership with government agencies, religious organizations and NGOs (ILO, 2014). This assistance included the provision of shelter, food, sanitary products, and medical and counselling services, as well as providing training on counselling to government workers. In addition, there have been awareness raising campaigns to prevent the stigmatization of returnees and discourage further irregular migration. Various programmes have aimed to empower returnees economically through entrepreneurship and vocational skills training and promoting access to finance. It ought to be noted, however, that the coverage of support was patchy (de Regt and Tafesse, 2016). In addition, ILO and NGOs have worked to increase the institutional capacity of government agencies, for instance developing a Reintegration Package with MoLSA (Kuschminder and Ricard Guay, 2017). Deportations from Saudi Arabia resumed in 2017; between 2017 and January 2019 approximately another 230,000 have been returned (Mixed Migration Centre, 2019).

METHODS AND DATA

Sampling and data collection

This paper is based on survey data that was collected on behalf of ILO Ethiopia and the Ethiopian Ministry of Labour and Social Affairs (MoLSA). In total, 2,039 returnees from Saudi Arabia were interviewed using structured questionnaires in June 2014. The majority of respondents were deportees involved in the mass deportations from Saudi Arabia to Ethiopia between the 13 November 2013 and the 7 December 2013. At the time of interview, most respondents would have been in Ethiopia for approximately six months.

A mixture of purposive and random sampling methods was used in order to ensure that the sample would be as widely representative as possible. The interviews took place in the Amhara, Oromia, Southern Nations Nationalities and Peoples Region (SNNPR), Tigray and the Addis Ababa City Administration regions, which together account for roughly 90 per cent of the population (Federal Democratic Republic of Ethiopia Central Statistical Agency, 2013). The sample size for each region was determined based on the numbers of returnees residing there. Zones, woredas (districts) and kebeles (wards) were selected purposively according to the concentration of returnees and ensuring the inclusion of both urban and rural settings. Returnees from the chosen kebeles were selected randomly from lists at woreda level to take part in the survey. The data collection also aimed to include at least 15 per cent of respondents from vulnerable groups (i.e. pregnant women, women with children under five and people with a disability). There is a near even spread of respondents from each of the five sampled regions. However, the gendered allocation of respondents is skewed to different locations, which is a limitation in the data; almost 40 per cent of female respondents were interviewed in Addis Ababa.

Variables

An overview of the variables is available in Table 1. The dependent variable concerns the returnees’ subjective evaluation of their return. The respondents were asked “At this point in time, how do you evaluate your return to your country of origin?” Those who responded “positively” or “very positively” were scored 1, whereas those who responded “neutral” or “negatively” were scored 0.

Table 1. Variables used for analysis

Description	Labels
Dependent Variable
At this point in time, how do you evaluate your return to your country of origin?	0 Neutral/Negatively 1 Positively/Very Positively
Independent Variables
Conditions prior to Migration
How would you describe your socio-economic situation before you migrated abroad?	1. Bad/ Worse 2. Medium 3. Good/ Very Good
*’Reason for migration’ was not used because over 95% of the sample responded positively to “seeking employment” or “seeking better life/ higher pay.”
Migration Episode
How would you describe your socio-economic situation abroad?	1. Bad/ Worse 2. Medium 3. Good/ Very Good
Have you acquired any specific professional or technical skills while you were abroad?	0 No 1 Yes
Did you face a challenge abroad? (assault, theft, rape, discrimination, verbal violence, abandoning wages, incompetence, culture shock)	0 No 1 Yes
Did you have acquaintances who lived there (in that country)?	0 No 1 Yes
Conditions upon Return
What is your employment situation since returning to your home country?	0 Unemployed 1 Self-Employed 2 Employed
Appropriate asset index score to reflect wealth in urban or rural context	-
Are you a member of one or more associations?	0 No 1 Yes
Did you receive support from friends, family, relatives or your community?	0 No 1 Yes
Support upon return
Did you receive support from a formal institution your reintegration? Government NGO IOM/ILO Health Institution	0 No 1 Yes
Control variables
Gender	1 Male 2 Female
Marital Status	0 Single 1 Married
Residing in an Urban Area	0 No 1 Yes
Education Level Unable to read and write (reference) Able to read and write/ First Second cycle Grade 5-8 Secondary School Grade 9-10 Preparatory/ 12 grade completed/ certificate or diploma	0 No 1 Yes
Number of children	-
How long did you stay in your country of destination? (years)	-

The independent variables are organized by stage of the migration cycle in order to support the analysis of different stages of the migration cycle on evaluation of return. For pre-migration conditions, we use the variable of socio-economic position before migration, which is measured subjectively as bad, medium or good. Under conditions during migration, we include variables encompassing both economic and social aspects: socio-economic situation abroad (measured subjectively), acquiring a skill abroad (binary), experienced a challenge abroad (binary) and having had an acquaintance abroad (binary). A challenge abroad includes experiencing one of; assault, theft, rape, discrimination, verbal violence, abandoning wages, incompetence or culture shock. We recognize that any of these experiences could influence mental health and well-being both while abroad and upon return. Variables for conditions upon return include current employment status, an asset index^1.1. The asset index was created to reflect the long-term economic status of the respondent using principal component analysis (PCA) in Stata. This method was completed using the guidance of the World Food Programme Vulnerability Analysis Mapping (World Food Programme, 2017). PCA provides a relative measure of wealth with greater weight given to factors that fewer respondents may have. As t-tests showed that there were significant differences at the one per cent level in the ownership of certain assets (e.g. livestock) between urban and rural populations, separate indices were created and assigned to the respondents depending on whether they had indicated that they were based in an urban or rural area.
, membership of an association (binary), and support from friends, family or the community (binary). Regarding support, it was asked whether the respondent received support from different institutions. Finally, the control variables include sex, marital status, urban or rural residence, education level, number of children and number of years abroad.

Analysis

The analysis was conducted in two parts. First, we examine descriptive level differences between those who evaluate their return “negatively/neutrally” or “positively/very positively” to gauge which factors appear to have the greatest impact upon evaluation of return. t-Tests were used to determine whether there were significant differences between the means of the different groups. Second, for the main empirical analysis, we first checked for collinearity. The independence of the variables was a particular concern, as it is expected that different parts of the migration cycle will impact the next. A correlation matrix was created comprising all variables, for which none of the included variables have a higher correlation of magnitude 0.32 and was thus deemed not problematic. Additionally, when building the model, we looked for collinearity by adding variables separately to see if they had adverse effects on the coefficients of other variables. We conducted logit regression analyses as the dependent variable is binary and it would give the clearest indication of what factors contribute to a positive evaluation of return. We ran the model having divided the variables into sub-groups to represent different stages of the migration cycle. In the final model, we include all the independent variables and control variables in order to test how they interact and affect the return experience. As will be explained further, a model that excludes employment status on return is also included as this variable may be capturing some of the effect of other variables.

RESULTS

The results are presented in two sections. First, we provide a descriptive analysis of the results, and second, the logistic regression results are discussed.

Descriptive Analysis

Examining the background characteristics, we see that the majority of respondents were male (56.3%) (Table 2). Most respondents reported that they were married (54.8%), living in a rural area (58.2%) and had no children (65.7%). Roughly 38 per cent of respondents were unable to read and write or only attended the first cycle of education (grades 1-4). The majority (58%) completed either the second cycle (grades 5-8) or secondary school entirely (grades 9-10). Only a very small minority (approx. 4%) attended school beyond grade 10. While the mean number of years spent abroad is only approximately three years and three months, ten per cent of respondents spent between eight and 26 years in Saudi Arabia.

For the dependent variable, it is surprising that although respondents were primarily deportees, a relatively large proportion (44.7%) evaluated their return as “positive” or “very positive,” while just over half of the respondents (55.2%) evaluated their return as “neutral” or “negative.”

It is notable in the descriptive statistics that socio-economic positions prior and during the migration episode appear to have opposite effects on evaluation of return. A minority of those who considered their socio-economic position before migration to be “good/very good” evaluated their return positively (38.2%). Conversely, just over half of those who described their socio-economic position to have been “bad/worse” (53.9%) evaluated their return positively. This would suggest that those who were economically worse off before migration evaluate their return more positively. There appears to be the opposite trend in evaluation of return for socio-economic position abroad; one-third of those who had a “bad/worse” socio-economic position abroad evaluated their return positively, whereas it is same for approximately half of those who had a “good/very good” socio-economic position abroad.

Overall, a much larger proportion considered their socio-economic position abroad to have been “good/very good” (61.1%) than prior to migration (27%). It can also be seen that many respondents (40%) acquired a professional or technical skill. Nevertheless, there do not seem to be major differences in evaluation of return by groups of those who acquired a skill, had an acquaintance abroad or faced a challenge. The challenges experienced abroad in order of frequency were as follows: language problems (71%), verbal violence (52%), incompetence (52%), abandoning wages (40%), discrimination (38%), cultural shock (33%), physical assault (23%), theft (22%), rape (4%) and other (1%). According to this survey, there were no significant differences between the abuses reported my male and female respondents. It is particularly surprising that for those who faced a challenge abroad, close to half have a positive evaluation of return.

Table 2. Descriptive statistics

Evaluation of Return	Negative/Neutral		Positive		Total
	No.	%	No.	%	No.	%
Dependent Variable
Evaluation of Return	1,119	55.23	907	44.77	2,026
Conditions Prior to Migration
Socio-Economic Position before Migration****** indicates statistically significant mean difference across groups at the 1% level
Bad/Worse	280	46.13	327	53.87	607	30.02
Medium	499	57.36	371	42.64	870	43.03
Good/Very Good	337	61.83	208	38.17	545	26.95
Migration Episode
Socio-Economic Position during Migration
Bad/Worse	236	66.11	121	33.89	357	17.65
Medium	239	55.45	192	44.55	431	21.30
Good/Very Good	642	51.98	593	48.02	1235	61.05
Acquired a Technical Skill
No	639	56.60	490	43.40	1129	60.09
Yes	406	54.13	344	45.87	750	39.91
Acquaintance Abroad **** indicates statistically significant mean difference across groups at the 5% level.
No	589	57.58	434	42.42	1023	51.69
Yes	507	53.03	449	46.97	956	48.31
Faced at Least One Challenge****** indicates statistically significant mean difference across groups at the 1% level
No	82	40.20	122	59.80	204	10.07
Yes	1037	56.95	784	43.05	1821	89.93
Conditions Upon Return
Employment Status after Migration****** indicates statistically significant mean difference across groups at the 1% level
Unemployed	664	50.42	653	49.58	1317	66.31
Self-employed	380	67.02	187	32.98	567	28.55
Employed	49	48.04	53	51.96	102	5.14
Asset Index ****** indicates statistically significant mean difference across groups at the 1% level
(mean)	0.18		-0.01		0.01
Support from Family, Friends, Relatives or Community
No	379	54.45	317	45.55	696	34.37
Yes	739	55.61	590	44.39	1329	65.63
Member of an Association****** indicates statistically significant mean difference across groups at the 1% level
No	767	52.53	693	47.47	1460	72.06
Yes	352	62.19	214	37.81	566	27.94
Support Upon Return
Support from Government ** indicates statistically significant mean difference across groups at the 10% level.
No	463	52.85	413	47.15	876	43.24
Yes	656	57.04	494	42.96	1150	56.76
Support from an NGO
No	741	55.34	598	44.66	1339	66.16
Yes	376	54.89	309	45.11	685	33.84
Support from ILO or IOM****** indicates statistically significant mean difference across groups at the 1% level
No	888	52.20	813	47.80	1701	84.04
Yes	229	70.90	94	29.10	323	15.96
Support from a Health Institution**** indicates statistically significant mean difference across groups at the 5% level.
No	1016	54.42	851	45.58	1867	92.24
Yes	101	64.33	56	35.67	157	7.76
Control Variables
Gender
Male	649	56.83	493	43.17	1142	56.37
Female	470	53.17	414	46.83	884	43.63
Marriage Status****** indicates statistically significant mean difference across groups at the 1% level
Single	468	51.49	441	48.51	909	45.16
Married	641	58.06	463	41.94	1104	54.84
Area of Residence
Rural	673	57.03	507	42.97	1180	58.24
Urban	446	52.72	400	47.28	846	41.76
Education****** indicates statistically significant mean difference across groups at the 1% level
Unable to read and write	189	62.58	113	37.42	302	14.91
Able to read and write/First Cycle 1-4	271	57.42	201	42.58	472	23.30
Second cycle Grade 5-8	403	54.83	332	45.17	735	36.28
Secondary School Grade 9-10	220	50.11	219	49.89	439	21.67
Preparatory/ 12 grade completed/ certificate or diploma	36	46.15	42	53.85	78	3.85
Number of Children****** indicates statistically significant mean difference across groups at the 1% level
(mean)	0.84		0.66		0.77
Years Spent Abroad ** indicates statistically significant mean difference across groups at the 10% level.
(mean)	3.18		3.40		3.28

• *** indicates statistically significant mean difference across groups at the 1% level
• ** indicates statistically significant mean difference across groups at the 5% level.
• * indicates statistically significant mean difference across groups at the 10% level.

Regarding the variables related to post-return conditions, it is striking that 66 per cent of the deportees sampled were unemployed approximately six months after their return. Those who were formally employed are a very small minority (approximately 5%). While roughly half of those who were unemployed described their return positively, only one-third of those who were self-employed responded the same. Perhaps most surprisingly, the mean asset index score for those who evaluated their return negatively is significantly higher than that of those who evaluated their return positively. The variables regarding social aspects of return do not show great differences in evaluation of return, with only a slightly lower rate of positive evaluations of return from those who are a member of an association.

Overall, 65 per cent of respondents received support from a formal institution and 35 per cent responded that they did not receive any support. Of those institutions, government agencies appear to have reached the most respondents (56.8%), followed by NGOs (33.8%), ILO or IOM (16%) and health institutions (7.8%). In most cases, the type of institution providing support does not appear to have affected evaluations of return greatly. An exception is that relatively fewer (29.1%) of those who received support from ILO or IOM described their return positively.

Main results

This section presents the main results of our logistic regressions. Table 3 displays the results of the logistic regressions as odds ratios. The following models follow a migration life-cycle structure. The final two models include all stages of the migration life cycle, but the final model excludes employment status post-return, as this variable appears to be correlated with gender, being a member of an association, receiving support from a government or health institution, and the number of years spent abroad, all of which have increased significance when this variable is excluded. Thus, it is excluded in the final model in order to examine the significance of these factors.

Table 3. Main analysis logistic regressions - odds ratios

Evaluation of Return	(1)	(2)	(3)	(4)	(5)	(6)	(6)
	Conditions prior to Migration	Migration Episode	Conditions upon Return	Support Upon Return	Control	Full Model	Full Model without Employment on Return
Conditions Prior to Migration
Socio-economic position before migration
1. Bad/ Worse	Ref.					Ref.	Ref.
2. Medium	0.637****** p < 0.01, **p < 0.05, *p < 0.1 (0.068)					0.624****** p < 0.01, **p < 0.05, *p < 0.1 (0.082)	0.617****** p < 0.01, **p < 0.05, *p < 0.1 (0.0794)
3. Good/ Very Good	0.528****** p < 0.01, **p < 0.05, *p < 0.1 (0.063)					0.551****** p < 0.01, **p < 0.05, *p < 0.1 (0.078)	0.569****** p < 0.01, **p < 0.05, *p < 0.1 (0.0800)
Migration Episode
Socio-economic position during migration
1. Bad/ Worse		Ref.				Ref.	Ref.
2. Medium		1.518****** p < 0.01, **p < 0.05, *p < 0.1 (0.238)				1.441****** p < 0.01, **p < 0.05, *p < 0.1 (0.249)	1.450****** p < 0.01, **p < 0.05, *p < 0.1 (0.247)
3. Good/ Very Good		1.715****** p < 0.01, **p < 0.05, *p < 0.1 (0.229)				2.112****** p < 0.01, **p < 0.05, *p < 0.1 (0.323)	2.126****** p < 0.01, **p < 0.05, *p < 0.1 (0.319)
Acquired a professional skill		1.120 (0.109)				1.337****** p < 0.01, **p < 0.05, *p < 0.1 (0.153)	1.330****** p < 0.01, **p < 0.05, *p < 0.1 (0.150)
Faced at least one challenge		0.574****** p < 0.01, **p < 0.05, *p < 0.1 (0.094)				0.783 (0.152)	0.747 (0.142)
An acquaintance in the country		1.163 (0.111)				1.130 (0.127)	1.150 (0.127)
Conditions Upon Return
Employment status after return
1. Unemployed			Ref.			Ref.
2. Self-Employed			0.533****** p < 0.01, **p < 0.05, *p < 0.1 (0.063)			0.563****** p < 0.01, **p < 0.05, *p < 0.1 (0.076)
3.Employed			1.243 (0.275)			1.055 (0.283)
Support from friends, family, relatives, community			0.978 (0.101)			1.021 (0.132)	1.050 (0.134)
Asset_Index			0.944 (0.038)			0.981 (0.044)	0.939 (0.0418)
Member of an association			0.877 (0.102)			0.838 (0.111)	0.804****** p < 0.01, **p < 0.05, *p < 0.1 (0.105)
Support Upon Return
Government institution				0.811****** p < 0.01, **p < 0.05, *p < 0.1 (0.078)		0.852 (0.102)	0.780****** p < 0.01, **p < 0.05, *p < 0.1 (0.0922)
NGO				1.047 (0.105)		1.065 (0.147)	1.073 (0.146)
IOM or ILO				0.380****** p < 0.01, **p < 0.05, *p < 0.1 (0.063)		0.362****** p < 0.01, **p < 0.05, *p < 0.1 (0.075)	0.350****** p < 0.01, **p < 0.05, *p < 0.1 (0.0729)
Health Institution				1.493****** p < 0.01, **p < 0.05, *p < 0.1 (0.342)		1.535 (0.429)	1.611* (0.448)
Control Variables
Female					1.110 (0.112)	1.197 (0.151)	1.264****** p < 0.01, **p < 0.05, *p < 0.1 (0.155)
Married					0.837****** p < 0.01, **p < 0.05, *p < 0.1 (0.084)	0.855 (0.100)	0.833 (0.0960)
Urban					1.041 (0.108)	0.972 (0.125)	0.988 (0.124)
Education Status
1. Unable to read and Write					Ref.	Ref.	Ref.
2. Able to read and write/ First Cycle 1-4					1.266 (0.195)	0.980 (0.184)	0.949 (0.177)
3. Second cycle Grade 5-8					1.301* (0.187)	1.056 (0.180)	1.046 (0.177)
4. Secondary School Grade 9-10					1.524****** p < 0.01, **p < 0.05, *p < 0.1 (0.242)	1.176 (0.220)	1.174 (0.217)
5. Preparatory/ 12 grade completed/ certificate or diploma					1.762****** p < 0.01, **p < 0.05, *p < 0.1 (0.465)	1.531 (0.489)	1.634 (0.521)
Number of Children					0.950 (0.035)	0.965 (0.043)	0.960 (0.0439)
Years abroad					1.030****** p < 0.01, **p < 0.05, *p < 0.1 (0.017)	1.028 (0.020)	1.034****** p < 0.01, **p < 0.05, *p < 0.1 (0.0201)
Constant	1.168* (0.095)	0.683 (0.160)	1.040 (0.097)	1.010 (0.076)	0.552****** p < 0.01, **p < 0.05, *p < 0.1 (0.113)	0.583 (0.223)	0.504****** p < 0.01, **p < 0.05, *p < 0.1 (0.186)
Observations	2,022	1,833	1,759	2,024	2,005	1,584	1,604

• Standard errors in parentheses
• *** p < 0.01, **p < 0.05, *p < 0.1

It is notable that a variable within each of our dimensions of interest is significant in the results. First, we see that socio-economic position before migration is significant across all models. Respondents that had “medium” or “good/very good” socio-economic position are on average much less likely to evaluate their return positively (on average a half to two-thirds as likely as a respondent who had a “bad/worse” socio-economic position).

Regarding conditions during migration, respondents that reported a “good/very good” socio-economic position during migration are significantly more likely to have a positive evaluation of their return across all models (on average over two times more likely than a person who was in a bad/worse socio-economic position abroad). Having acquired a professional skill while abroad also increases the likelihood of a more positive evaluation of return in the models including all variables.

In the penultimate model, the only variable in post-return conditions that is significant is employment status upon return, where respondents that are self-employed are significantly less likely to report a positive evaluation of their return. When this is removed, being a member of an association is significant at the ten per cent level, which suggests that part of the effect of these variables are explained by the employment variable. Those who are part of an association might be slightly less likely to evaluate their return positively.

When examining the support received, support from the ILO or IOM is highly significant across all models. On average, respondents who received support from the ILO or IOM appear to be three times less likely to evaluate their return positively than those who did not receive it. Receiving support from government or a health institution is significant at the five and ten per cent levels in the final model, suggesting that those who received support from government are slightly less likely to evaluate their return positively, whereas those who received support from health institutions are 50 per cent more likely to evaluate their return positively.

Background characteristics, including marital status, number of children and urban residence, are unexpectedly not significant across all models. The only background characteristic that is initially significant appears to be if the respondent is highly educated. According to Model 5, respondents appear to be more likely to evaluate their return positively if they have a higher level of education, but this is not significant in the final models. On average, it is suggested in the final model, at the ten per cent significance level, that female respondents and those who spent more years abroad might be slightly more likely to evaluate their return positively.

DISCUSSION

The results show that each stage of the migration life cycle has significance in the evaluation of return, illustrating the importance of Cassarino’s preparedness model. First, we see that the socio-economic position prior to migration strongly influences the evaluation of return and that this occurs in a reverse relationship from our hypothesis (H1). That is, respondents with a better socio-economic position prior to migration are more likely to have a negative evaluation of their return. One reason for this could be that upon return the respondents were unable to achieve their pre-migration conditions. Descriptive statistics show that 35.9 per cent of respondents considered their socio-economic position after migration to be worse than prior to the migration. The inability to prepare for the migration by gathering resources from the destination country thus has a strong negative impact on the reintegration, reasserting preparedness theory.

Second, regarding experiences in Saudi Arabia, socio-economic position is also highly significant, though in such a way that our hypothesis is incorrect (H2); respondents with a better socio-economic position abroad are more likely to have a positive evaluation of their return. Even though the migration episode was interrupted, these respondents may still have had experiences that helped them in their return, for instance gaining technical skills or living independently. Additionally, those with a better socio-economic position abroad may have had some more resources in hand for their return, though not in such large amounts that it shows in the effect of the asset index variable. As the expulsions from Saudi Arabia progressed, deportees may have been better prepared for their departure. Another potential inference is that, due to their success abroad, these respondents were able to return with pride and dignity despite being expelled, which may also have contributed to their positive evaluations of their return.

Similarly, respondents that acquired a professional skill abroad also had a more positive evaluation of their return (H4), and we assume this could be for the same reasons mentioned above. Interestingly, the effect of acquiring a skill abroad is only significant in the final models that include variables from all stages of the migration cycle. This supports Collyer’s (2018) finding that the positive development effects of human capital acquired abroad are dependent on the wider context of return.

It is surprising that facing a challenge abroad was not significant in the final model (H3): however, we expect that this is because of the high prevalence of respondents that faced a challenge (90%) and the resulting low level of variation in this variable. This descriptive finding, however, provides further evidence and concern for the treatment of migrant workers in Saudi Arabia.

For conditions upon return, it is highly significant that those who were self-employed were much less likely to have a positive evaluation of their return. One reason for this is that self-employment is engaged in as a form of necessity entrepreneurship. A limitation of this study is that we do not assess how the self-employment activity is performing, so these respondents could be struggling. Unexpectedly, the asset index is not significant across all models. Although it is expected that a higher asset index score would influence a positive evaluation of return, there appears to be a negative relationship, possibly because those who are self-employed own the most assets. The mean asset index score of those who are self-employed is 0.61, compared to 0.12 for those who are employed, and −0.11 for those who are unemployed. Therefore, as those who were self-employed evaluated their return more negatively, this influences the effect of the asset index. In addition, as 71.3 per cent of those who are self-employed live in rural areas, one may infer that some of the challenges of reintegration may be linked with the precarity of agricultural work. Indeed, this study raises important questions regarding work conditions and government support for both those who are employed and self-employed in Ethiopia.

Fourth, receiving support was significant in the results; respondents that received support from either IOM or ILO reported to have more negative experiences in return. In this instance, the programmes were targeted at vulnerable deportees who one might expect to struggle on their return. Thus, this result suggests alignment in the programme aims and the profile of respondents who received support.

Across the results of all stages of the migration cycle, it is surprising that networks and social reintegration indicators were not significant in the analysis. This could be due to the fact that first, networks have limited capacity to offer support in the Middle East (Kuschminder, 2016), and second, that the main motivation for the migration was economic; thus, it is consistent that economic variables have more significance. In other country contexts, however, social and network variables may have more significance.

A second surprising finding was that gender was not significant in the results. This contradicts other studies that have found highly significant results for gendered experiences of return to Ethiopia (Bilgili et al., 2018). One reason for this is that this study focuses on the singular case of mass deportation from Saudi Arabia, where the other study focused on different forms of return. The gendered result in this study stresses that in the case of mass deportation from Saudi Arabia to Ethiopia vulnerability should not be assessed based on gender. Interviews in Ethiopia with key stakeholders stressed this concern that male returnees were often excluded from being considered as vulnerable due to their gender. There has been recognition from front-line workers in Ethiopia that both male and female returnees experience vulnerabilities upon return. This is one of the first studies of reintegration in Ethiopia that has fairly equal gender respondents as comparative studies have higher responses from female returnees (De Regt and Tafesse, 2016; Bilgili et al., 2018).

CONCLUSION

In terms of short-term reintegration experiences, the findings illustrate that employment continues to be a central challenge for the returnees. Only five per cent of respondents were employed. Those in self-employment seem to be engaged in necessity entrepreneurship (29%) and the majority were unemployed (66%). Comparatively, 66 per cent reported receiving support from family, friends or their community, and 28 per cent were part of an association post-return. The findings illustrate that economic concerns are dominant in Ethiopia. The Government of Ethiopia, international organizations and NGOs have been engaging in significant effort to increase employment opportunities and economic reintegration outcomes for this group, which the evidence demonstrates is well aligned to needs. They could further focus on ensuring that work conditions are such that returnees can sustain a livelihood, particularly in rural areas. In many cases, poor working conditions with low pay motivates labour migration to the Middle East.

Our findings demonstrate that evaluations of return within a context of mass deportation from Saudi Arabia to Ethiopia are more positive than would be expected considering the context and literature on deportation. This finding contrasts much of the deportation literature and challenges our understanding of reintegration after deportation (for example Khosravi, 2018). We consider three possible explanations for this finding. First, we must consider the specificities of this migration flow. The significant lack of job opportunities in Ethiopia is a key factor driving this flow, despite the conditions abroad. Having a good socio-economic position while abroad was significant in the regression model and illustrates the importance of employment possibilities for Ethiopians. Therefore, having had the opportunity for employment abroad may reflect in this case in having a positive return.

Second, and related to the above, we regard this positive experience of return as possibly influenced by poor conditions abroad; 90 per cent of respondents experienced challenges abroad, which may have resulted in too low variation in the model to be significant. Other research overwhelmingly demonstrates the significance of this point and it must be considered as central to the explanation for the positive evaluation of return (Fernandez, 2013; Kuschminder 2017).

Finally, there is also the possibility that although deported with an interrupted migration cycle, some respondents may have wanted to return. In a study on return migration to Ethiopia from the Middle East, Kuschminder (2017) found that women purposively identified themselves to police in order to be deported. This was done when they were in abusive situations with their employers and could not afford to return to Ethiopia on their own. This situation was termed as “fighting to flee” and represents a paradox in the forced/voluntary return dichotomy as they women are placed in the deportee box as forced returnees, but actually choose for deportation and voluntarily wanted to return. The possibility of this situation is unclear within this quantitative data, but these qualitative insights allow for another possibility to explain the results.

On the whole, the findings suggest expanding notions of the forced versus voluntary return dichotomy as has been argued by Erdal & Oeppen 2018, p. 993) wherein we need more reflection on “the complexity of individual experiences, agency and contextual circumstances than a binary labelling of forced or not.” This case and article show that deportation is not always a negative return experience, which is contrary to expectation. This is a critical area for further investigation and understanding if this is the case in other countries, such as mass deportations from Saudi Arabia to Pakistan or Bangladesh. Experiences in mass deportations are an area for further investigation in other country contexts, and thus, it would be informative to have comparative assessments of deportees from other contexts. The conditions in these mass flows may challenge our existing understandings, create new policy challenges and create vulnerabilities due to the scale of return that require further research and understanding.

In Ethiopia, the number of deportees from Saudi Arabia has increased in recent years and this analysis contributes to informing policy and programming on the experiences and needs of deportees upon return. Ethiopia has made significant progress in the governance of reintegration and in 2018 passed a proclamation for reintegration. The proclamation declares the rights of returnees and provides a framework for assisting returnees in their reintegration (Kuschminder and Ricard Guay, 2017). Implementation has been stalled due to shifting priorities with increasing refugees and internally displaced persons in Ethiopia (Ogahara and Kuschminder, 2019). A central recommendation for this policy is to develop a case management system for providing comprehensive reintegration support across the country. This would include reintegration extension officers that meet returnees in their local communities to discuss their reintegration and connect them with necessary services. Such a policy is a large endeavour but would ensure that all returnees are able to have information on reintegration and access necessary services for support, in particular for the most vulnerable that may otherwise be left behind. Finally, further research is required on longer term reintegration, for which longitudinal studies tracking returnees would be highly informative in understanding their reintegration processes and needs.

Note

1. The asset index was created to reflect the long-term economic status of the respondent using principal component analysis (PCA) in Stata. This method was completed using the guidance of the World Food Programme Vulnerability Analysis Mapping (World Food Programme, 2017). PCA provides a relative measure of wealth with greater weight given to factors that fewer respondents may have. As t-tests showed that there were significant differences at the one per cent level in the ownership of certain assets (e.g. livestock) between urban and rural populations, separate indices were created and assigned to the respondents depending on whether they had indicated that they were based in an urban or rural area.