Return migration: Evidence from a reception country with a short migration history

Abstract

Much discussion can be found in the literature on the determinants of immigrants’ decisions to repatriate. Yet, missing is the identification of such determinants where the host country has become a reception country relatively recently. The main objective of this paper is to identify the determinants influencing the decisions of Albanian migrants in Greece to return home. Research is based on quantitative analysis techniques applied to a survey sample of 200 Albanian migrants returning from Greece to Albania. Results indicate that the main determinants driving them back home include the difficulty of integrating into Greek society, racism and failure to find work. Despite certain limitations of the study, the unconventional return migration determinants identified may become key considerations for migration policy makers for many related policy issues in cases where the reception country has a short history of migration.

Keywords

Categorical models reception country return migration causes short migration history

Introduction

The presence of immigrants in Greece constitutes a new and very important issue for the economic and social life of the country. A considerable number of Albanian immigrants entered over a short period of time, causing a shock to the country that affected the society and its economy and forming a significant and novel situation. The immigration phenomenon, at least during the first decade, was confronted with disdain and a number of impromptu measures. The policy measures of that period, whether taken seriously or not, are probably directly related with the reasons that forced a number of Albanian immigrants to return home.

Though Greek immigration has a long history all over the world, Greece as a reception country has a very short migration history; it became a host country during the last two decades (see, among others, Labrianidis and Kazazi, 2006; Lazaridis and Koumandraki, 2001; Lazaridis and Psimmenos, 2000; Rovolis and Tragaki, 2006). Greece’s migration history is much shorter than that of any other country of Southern Europe (Italy, Spain, and Portugal) that began accepting immigrants in the 1980s (Massey et al., 1993). A substantial number of Albanian immigrants came to Greece, ranking it as the main destination country for Albanian emigrants. As mentioned in Labrianidis and Kazazi (2006), the total number of Albanians in Greece is estimated to be between 400,000 and 550,000, while Vadean and Piracha (2010) raise this number to 600,000.

As a new phenomenon in Greece, immigration has only recently occupied the interest of research experts and policy makers. Kasimis (2008) examines the impact in the agricultural sector, Labrianidis and Sykas (2009) examine the effects on rural areas and Lazaridis and Koumandraki (2003) examine the issues of ethnic entrepreneurship. Among others, Piracha and Vadean (2010) examine issues that relate economic effects and remittances, and whether or not return migration and remittances have important potential in promoting growth and development. Nikas and Baklavas (2009) analyse the relationship between, on the one hand, savings and investment attitudes of Albanian emigrants and, on the other, migrants’ personal characteristics and geographic origin.

The presentation and evaluation of issues related to Greek policies on migration and integration of Albanian immigrants within Greek society are other issues addressed by the literature. Fakiolas and King (1996) review aspects of Greek migration policy and comment on difficulties of formulating migration policy and factors that may lead to serious difficulties in terms of immigrants’ integration. Fakiolas, in two other studies (Fakiolas, 2003a, 2003b), examines regularisation and legalisation of undocumented immigrants in Greece; analogous issues were examined by Lazaridis and Poyago-Theotoky (1999); Glytsos (2005); Iosifides et al. (2007); Triandafyllidou (2000, 2009).

Social exclusion and racism problems that emerged after the massive inflow of Albanian immigrants, and which influence return migration, were also examined. Lazaridis and Psimmenos (2000) investigated the social and economic exclusion of Albanian immigrants in Greece. Lazaridis and Koumandraki (2001) presented a thorough analysis on the existence of racism in Greece and how it is experienced by ethnic minorities. They stressed the shift in colour racism as the primary indicator of social exclusion and describe how migration has induced a number of negative reactions among Greeks. Lazaridis and Koumandraki (2007) also provided a thorough analysis on social exclusion with reference to Albanian migrants in Greece. Finally, the study by Triandafyllidou (2007) investigated racism and social exclusion of immigrants in Greece and Italy.

Almost all studies that deal with the aforementioned issues highlighted the many problems and deficiencies of Greek migration policies and regularisation programmes, and stressed the short migration experience of Greek authorities. As mentioned by Triandafyllidou (2007), only in 1991 did Greece introduce the first legislative framework for controlling and managing immigration. Of course, these deficiencies affected and shaped the behaviour of the local population in a way that made the incorporation and social inclusion of the immigrants more difficult. The adaptation of immigrants in the host country depends on a number of factors (Kosic and Triandafyllidou, 2003). Such factors are immigrant sociodemographic attributes, their support networks and the institutional environment of the host country. Thus, not only employment opportunities, but also policy implementation practices, welfare benefits and overall socioeconomic integration determine the degree of immigrant adaptation in the host country.

Yet, the issue that gained the least attention, and which has not yet been studied, is that of return migration. This can be partially explained by Greece’s short history as a host country, beginning only in the 1990s, when large-scale immigration transformed Greece from an emigration country. However, many Albanian immigrants have started to return home. So far, there has been little analysis of return migration from countries with a short migration history; a gap this paper seeks to fill by examining the major reasons that drive Albanian immigrants to return back to their country, paying attention to their heterogeneity.

In order to proceed with the examination of the return migration phenomenon, a survey was conducted to monitor issues related to return migration reasons. In particular, part of the survey was designed to elicit data on immigrants’ return decision and to register their views on attributive return migration elements. The conducted research, based upon personal interviews with 200 participants, focused on 17 Albanian areas in the south and central part of the country during June and August 2007.

The rest of the paper is organised as follows. Issues affecting return migration are analysed in the next section. The third section outlines the research design and methodology. In the last two sections the results and concluding remarks are reported, respectively.

Return migration

Migration is considered as a form of change that can lead to transformations in both the sending and receiving countries (Portes, 2010). Accordingly, return migration, either cyclical or permanent, as part of the migration cycle, can cause various changes and may be induced by many determinants. Return migration is an integral part of any out-migration process (Zhao, 2002); it is one of the three stages of the migration cycle (King, 1986; Labrianidis and Hatziprokopiou, 2003). The three stages of the cycle are prior to, during and following migrant returns. As mentioned by Cassarino (2004), while some scholarly approaches related to return migration can be found from the 1960s, in the 1980s return migration stimulated the debate among scientific scholars in assessing its impacts. Since then, return migration studies have become one of the most researched subjects in international literature (Arango, 2000; Cassarino, 2004; Massey et al., 1993; Portes, 2010). Return migrants from the USA, during that period, were those that mainly initiated the interest of researchers to be engaged with the issue (King, 1986).

Scholars internationally examine a relatively broad set of subjects related to the issue of return migration, such as determinants of return, various impacts of return on both the reception and origin country, examination of the social and economic status of return migrants, the role of social networks, and the social inclusion status and migration policies. Despite the many studies, there is a lack of an integrated theory on return migration based on uniform theoretical justification; thorough analyses on the issue, and on return migration generally, can be found in Arango (2000), Cassarino (2004), Ghosh (2000), King (2000), Massey et al. (1993) and Portes (2010). As part of the migration cycle, return migration has been the subject of various migration approaches and theoretical models. Although each of those schools of thought sought to explain similar problems, they employed various different assumptions and references (Massey et al., 1993).

The set of principal schools of thought primarily consist of the approach of neoclassical economics, which mainly focuses on the differences in wages and employment between the sending and hosting countries. As mentioned by Portes (2010), it is based on an individualistic calculus of benefits and costs (income maximisation) among migrants (Borjas, 2001; Todaro, 1969). The new economics of migration is the next approach; as above, following Portes (2010), the ideas of this school are based on family strategies to overcome market imperfections in the sending countries. Migration is viewed as a family decision to maximise family income and not individual income (Massey, 1990; Stark, 1991; Stark and Bloom, 1985). World systems theory is another approach that does not pay attention to micro-level decision processes but focuses on macro-social processes and on the fact that less developed countries need labour with low wages in certain sectors (Portes and Walton, 1981; Sassen, 1988).

The structural approach supports the idea that migrant return depends not only on the individual experience of the migrant but also on social and institutional factors in the origin country. The crucial role of financial and economic resources brought back to the country of origin is stressed in the decision to return. Cerase (1974) identified four different types of returnee from the relationship between the return migrants’ expectations and the social and economic situation found in the home country; they are the returns due to failure, conservatism, retirement and innovation. Finally, the social and economic networks approach suggests that, in view of the last scholarly approach, migration networks are defined as relations among persons that consist of the nexus between migrants with relatives, friends or fellow countrymen (Castles, 2004). According to Massey et al. (1987), migration networks can be seen as a form of social change. The views on migration and return migration in the aforementioned scholarly approaches are discussed in the studies of Arango (2000), Cassarino (2004), Massey et al. (1993) and Portes (2010).

As reported by Cassarino (2004), those schools of thought include return migration as a subcomponent of their analysis, as an integrated theory on return migration does not exist. The examination of the causes of return migration, whether forced, planned or unplanned, temporary or permanent, is one of the many issues that has drawn the attention of scholars. Among the various causes that determine the decision of an immigrant to return, many can be explained by factors stemming from rational choices. Theoretically, the decision of a potential return can be based on economic factors such as social capital and cost–benefit evaluation.

On a macro-economic level, the difference in wages between the origin and the reception country is a decisive factor that causes the move. Workers with low wages decide more easily to return to their country of origin than workers who enjoy high wages in the reception country. On a micro-economic level, one can claim that an immigrant is more likely to return to his country if he is unemployed or if his job agreement tends to be at its end and unemployment allowances are limited (Massey and Constant, 2001). Furthermore, the social capital of an individual should be taken into consideration as it can also affect, positively or negatively, his or her decision to return. Specifically, the tendency to return is higher for individuals with limited social capital than individuals who managed to achieve a high level of social capital. This provides them with important benefits and social status in the reception country.

Stark (1997) placed family income at the centre of his/her analysis, with family income, rather than personal income, deemed to determine the decision to return. The role of social capital level in influencing migrant decisions to return home is questioned by models that examine variables such as immigrant social integration status and relations between their family and relatives. Results indicate that the existence of strong family relations in the reception country, the existence of effective social networks among them, as well as the availability of opportunities to develop their skills and knowledge, limit significantly the possibilities to return to their home country (Haug, 2001). The age of immigrants is an important variable with a strong influence in the decision to return. Results indicated that the likelihood to return is very low for people who are in the productive period of their lives (middle age), though this hypothesis is significantly affected by the type of job and the level of wages the immigrant enjoys. Much less is the likelihood to return for young (below 20 years) and for very old people. On the other hand, the highest rates of return migration were observed at retirement age (Moeser, 2005).

In their analysis, Vadean and Piracha (2010) present a number of arguments to explain the decision to return: location preferences; purchasing power and more favourable returns of accumulated capital in the home country; failure to achieve initial migration targets (not finding any job or the expected job). An explicit analysis of return migration causes can also be found in Dustmann (2003).

Albanian return migrants from Greece

Although several systematic studies have been conducted to examine issues related to migration in Greece, there is a lack of related research for the return migration phenomenon. The lack of such studies might be a result of the short-term presence of Albanian immigrants in Greece, with the immigrants only having arrived within the last two decades, making the return migration phenomenon very new. Paucity of data is another important reason stressed by Labrianidis and Hatziprokopiou (2003), Labrianidis and Kazazi (2006) and Piracha and Vadean (2010).

Other issues, such as impacts of immigrants on the economy, in agriculture, in employment, in criminality and legalisation situation, were extensively studied by many scholars, as discussed in the introductory section (Arapoglou and Sayas, 2009; Coulon and Piracha, 2005; Hatziprokopiou, 2004; King, 2004; Labrianidis and Sykas, 2009; Mai, 2001; Papapanagos and Sanfey, 2001). Many of these studies have found interesting results; however, their merits vary. Promising results, by applying rigorous research selection criteria, have been found in four recent studies focused on return migration of Albanians, who number the vast majority of immigrants in Greece (Labrianidis and Hatziprokopiou, 2003; Labrianidis and Kazazi, 2006; Labrianidis and Lyberaki, 2004; Vadean and Piracha, 2010).

The above four studies examine issues of numbers of Albanian immigrants to, and their return migration from, Greece and Italy along with a comprehensive presentation of the related literature. The study by Labrianidis and Hatziprokopiou (2003) examines the role of Albanian returnees in the economic development of their origin areas. Results indicate that Albanian immigrants return in a better economic situation. The study by Labrianidis and Lyberaki (2004) indicates that Albanian immigrants who have returned from Greece seem to be better equipped and adjusted to their host society and labour market than those returning from Italy, despite the fact that Greece attracts the lesser skilled and is not migrants’ first choice. The study by Labrianidis and Kazazi (2006) examines the role of return in regional disparities, specifically if migrants return to their place of origin or migrate internally and cause spatial impacts. By examining the relation between migration and internal migration support, the idea that return migration does not exacerbate spatial disparities is investigated. The most recently published study (Vadean and Piracha, 2010) examines, comparatively, the socioeconomic characteristics of circular and permanent return migrants.

Though the three studies refer to the reasons of return migration, it is observed that an expedient and purpose-built model analysis for the determinants of return migration is not present; it was not the main objective of any of the four studies. On the contrary, by further modelling the answers of a survey performed in Albania to a sample of return migrants, interesting results emerged in the present study. By clustering the return migrants, using a two-step cluster analysis, it was revealed that the dominant reasons driving Albanian immigrants to return were not positive economic reasons and family issues, as usual. The different results found in the current study constitute its main contribution; those results and the employed methodology are presented, analytically, in the next sections.

Methodology

Empirical analysis is based on recent fieldwork conducted in 2007 with interviews and qualitative questionnaires addressed to 40 recent return migrants to Albania who were not included in the final sample. This focus group comprised mainly key stakeholders in Albania (presidents of chambers, entrepreneurs, opinion leaders, etc.). The results of this qualitative approach were then used to design the final research questionnaire of the study.

According to Labrianidis and Kazazi (2006), the definition of permanent return migration includes the condition of returning and living in the country of origin for at least one year. However, in the current study, the qualitative research indicated that it is very difficult to find permanent Albanian returnees according to this definition –few of them return to Albania and live there for more than two months. In addition, seasonal or circular migration phenomena are very common in Albanian reality (Carletto et al., 2006; King and Vullnetari, 2003) owing to proximity and, as a result, the permanent character of any return migration decision is rather uncertain. Thus, in order to secure a representative sample (Labrianidis and Kazazi, 2006), we selected our questionnaire sample from among people who satisfied the following criteria: (a) they were over 18 years old when the survey was conducted; (b) they had stayed for more than one year abroad as emigrants; (c) they had decided to return to Albania and live there permanently; (d) they had returned to Albania and lived there for at least two months at the time of the interview; and (e) they had been emigrants to Greece. As the total number of Albanian returnees remains unknown and since much of the migratory movement, especially between Albania and Greece, has to do with flows back and forth, the only way to ensure a random sample was to make use of the snowball sampling method (Salganik and Heckathorn, 2004). This sampling technique is a non-probability sampling technique and is often used in hidden populations, which are difficult for researchers to access.

The empirical research was conducted by using aided self-administrated questionnaires (Malhorta, 1996). Data were collected during the period June–August 2007, through a survey addressing return migrants from 17 different areas of the Albanian land. According to the snowball sampling theory, participants were selected one by one, mainly from the following areas (Figure 1): Korçë, Elmpasan, Tiranë, Sarandë, Gjirokastër and Pogradec. In selecting members of the sample, care was taken to ensure that they originated geographically from a variety of different areas of Albania (north, centre, south) and from settlements of different population size (Tiranë, other urban areas, villages).

Figure 1.

Map of the selected research areas.

The above methods of selection resulted in a sample of 121 men and 79 women who have lived as migrants in Greece for at least one year, and returned permanently to Albania for more than two months at the time of the interview. However, as sample members are not selected from a sampling frame, snowball samples are subject to numerous biases (e.g. people who have many friends are more likely to be recruited into the sample). Thus, taking into account the existence of possible biases in selection of the sample, as well as definition vagueness, it would be rather dangerous to attempt strict and sweeping generalisations.

The final survey was designed to monitor issues related to return migration causes based on the qualitative research and existing literature (King, 1986; Labrianidis and Hatziprokopiou, 2003; Labrianidis and Kazazi, 2006; Patiniotis, 1985). In particular, part of the survey was designed to elicit data on migrants’ return and their views on attributive return migration elements. Thus, respondents were asked to indicate the main return migration reasons and provide several personal details to analyse how return migration was influenced by personal characteristics of the respondents and how this conforms to the general European or global practices. To encourage participation and to minimise the cognitive burden on respondents, most questions were framed using Likert scale intervals. Finally, the survey instruments were designed, tested and redeveloped according to the specificities of the research area.

In this paper, both summary statistics and multivariate analysis techniques were employed using the Statistical Package for the Social Sciences – version 17 (SPSS, 2007): two-step cluster analysis (TSCA) and categorical regression (CATREG). Although the methodology of the chosen empirical techniques is rather unusual, they have been selected because of their ability to handle categorical variables optimally (Charatsari et al., 2011; Michailidis et al., 2010). Indeed, many of the data that social and political scientists deal with are qualitative in nature and many other data are at best ordinal (Berry and Lewis-Beck, 1986). Political scientists often analyse data as though they meet the criteria for an interval scale, even though they fail to meet the assumption required for the methods used. However, with the categorical methods available for analysing qualitative data, one does not need to make the types of questionable assumptions that are often made.

In order to explore the different reasons for return migration and to classify return migrants in discernible clusters, with similar return migration behaviour, the TSCA was first employed as a scalable cluster analysis algorithm designed to handle large datasets, revealing natural groupings within a data set that would, otherwise, not be apparent (Siardos, 2002). Traditional clustering methods are considered effective and accurate on small data sets and, usually, do not scale up well to large data sets unless these data sets are first reduced into smaller ones. Moreover, traditional clustering methods cannot handle categorical variables or attributes most commonly found in sociological research surveys optimally (Zhang et al., 1996). Although TSCA requires only one data set, it follows a two-step procedure: the first step pre-clusters the cases into many small subclusters, and the second step clusters the subclusters of the first step into the desired number of clusters. The algorithm can also automatically select the number of clusters. As the number of subclusters is smaller than the number of original records, traditional clustering methods can be used effectively (SPSS, 2007).

Additionally, CATREG (Van der Kooij and Meulman, 1997) has been used in order to highlight possible relations between return migration decision and a set of other selected independent categorical variables. In fact, CATREG (one of the recent options in SPSS v. 17) is a modern regression technique, much more holistic and effective than the multiple regression analysis and the usually employed logit–probit models. Actually, both logit and probit models in logistic regression are special cases of a link function in a generalised linear model: they are the canonical link functions for the binomial distribution. On the other hand, the CATREG model can deal more optimally with both qualitative and quantitative data, as it works on two discrete and simple stages: first, the nominal and ordinal variables are transformed to interval scales, in order to maximise the relationship between each predictor and the dependent variable and, second, multiple regression analysis is applied to the transformed variables (SPSS, 2007). CATREG tools provide the framework for choosing between reference cell and effect cell parameterisation. This means that categorical variables, or interaction terms that include categorical variables, will drop or add the entire variable or interaction term and evaluate changes in model fit, rather than dropping one categorical level at a time. Comparatively, even though CATREG is relatively complicated and sophisticated, involving advanced statistical techniques such as optimal scaling techniques for multivariate categorical data analysis, there are several advantages in using this model. The main advantage is that CATREG can be run with the least assumptions: (a) the normality assumption of the predictor variables is relaxed, (b) factor levels are coded simultaneously into values, and therefore sample sizes need not necessarily be large, (c) only one coefficient is needed for a predictor variable and (d) non-linear associations can be detected with these models.

Relative importance indicates the importance of each predictor, using Pratt’s measure (Pratt, 1987). This measure is roughly equivalent to the product of the regression coefficient and zero-order correlation. The Pratt index is primarily used to uncover suppressor variables. That is, in the case of a predictor yielding a relatively high beta, but low importance, the situation suggests that the variable may have been suppressed by other predictors. In addition, partial and part correlations are similar to zero-order correlations, except that the effect of all other predictors has been controlled. Finally, tolerance is utilised to identify multicollinearity. According to econometrics literature (Pratt, 1987; Siardos, 2002; SPSS, 2007) relative importance measures are much more useful than the usual standardised beta weights. In particular, relative importance indicates the percentage of explanation of the dependent variable while also being used to predict the future values of the dependent one.

Data analysis and results

Table 1 presents a short description of the research sample. According to the data presented in the table, the representative participant of the study is male, 40 years old, married with two children, has higher education and stayed in Greece for 7.6 years. The majority of the respondents (64.0%) come from six specific areas in the southern and central parts of Albania. In particular, 15.5% of the respondents come from Pogradec, following by Korçë (13.4%), Gjirokastër (9.6%), Tiranë (8.8%), Sarandë (8.5%) and Elmpasan (8.2%). However, there is a considerable percentage of respondents, more than one-third of the sample, who come from 11 different areas.

Table 1.

Description of the sample.

Personal characteristics
Male (%)	60.5
Average age (years)	39.2
Average number of children	1.9
Married (%)	76.0
University degree (%)	38.0
Average number of years in Greece	7.6
Area
Korçë (%)	13.4
Elmpasan (%)	8.2
Tiranë (%)	8.8
Sarandë (%)	8.5
Gjirokastër (%)	9.6
Pogradec (%)	15.5
Other areas in the north (%)	36.0

TSCA, based upon the agreement level of the 11 return migration reasons by return migrants (5=strongly agree, 4=agree, 3=neither agree nor disagree, 2=disagree and 1=strongly disagree), was first used to classify the respondents in discernible clusters in order to explore the different reasons for return migration.

SPSS v. 17 for Windows was employed for the multivariate statistical analysis of the data set (200 cases). The two-step cluster method extracted automatically the optimal solution of three clusters. The majority of the respondents (71 or 35.5%) were included in the first and the third clusters, whereas only 58 (29.0%) were included in the second. Regarding the distribution of observations in the above clusters, depending on the reasons for return migration, it is shown in Table 2 that the second cluster constitutes mainly resettling immigrants who returned to their country of origin for family reasons (57.9%) or because of retirement (56.5%). On the other hand, the third cluster constitutes immigrants who returned to their homeland because they accomplished the gathering of enough money (60.0%) for their children’s education (59.6%) or for other reasons (76.7%), including (i) encouragement to return by friends and relatives, (ii) their return was actually planned from the start, (iii) achievement of their initial aims and (iv) some secondary psychological reasons. Finally, the first cluster is differentiated completely from the rest and it mainly includes immigrants who faced problems of racism (92.9%), failed to create their own work in Greece (74.1%), wished to establish an activity in Albania (68.9%), faced problems of incorporation in Greek society (61.1%), were unemployed (48.9%) or were nostalgic for their country (38.8%).

Table 2.

Distribution of observations in each cluster (frequencies and percentages).

Continuous variables	Clusters^a
	First (71, 35.5%)	Second (58, 29.0%)	Third (71, 35.5%)
1. Family reasons		58/100 (57.9%)
2. Nostalgia	19/49 (38.8%)^b
3. Incorporation problems in Greek society	11/18 (61.1%)
4. No constant work	23/47 (48.9%)
5. Wish to establish an activity in Albania	51/74 (68.9%)
6. Retirement		13/23 (56.5%)
7. Children’s education			28/47 (59.6%)
8. Racism problems	26/28 (92.9%)
9. Accomplished gathering enough money			48/80 (60.0%)
10. Failure to create their own work in Greece	20/27 (74.1%)
11. Other reasons			3/4 (75.0%)

First cluster: ‘unsuccessful migrants’; second cluster: ‘family priority immigrants or retired ones’; third cluster: ‘younger and successful immigrants’.

Figures in parentheses represent the proportion of observation of each variable in each cluster per the total observations of the variable in the sample.

In tabloid form, the return migrants of the first cluster can be defined as ‘unsuccessful migrants’, those of the second cluster as ‘family priority or retired immigrants’ and those of the third cluster as ‘younger and successful immigrants’. Table 3 highlights some statistical differences in the members’ profile of each cluster, which also support the aforementioned typology. Thus, the ‘unsuccessful migrants’ of the first cluster have the lowest education level. This first cluster has the highest percentage of married members. The ‘family priority or retired immigrants’ of the second cluster are older, with more children and have stayed in Greece longer than the other clusters. This cluster has the highest percentage of male members. Finally, the ‘younger and successful immigrants’ of the third cluster are completely differentiated from the returnees of the previous clusters: they have fewer children, have stayed in Greece for less time, are younger and better educated. This cluster has the lowest percentage of both married and male members.

Table 3.

Members’ profile of each cluster.

Personal characteristics	First cluster	Second cluster	Third cluster
Male (%)	62.0	67.2	53.5
Average age	40.3	43.2	34.8
Average number of children	2.1	2.2	1.4
Married (%)	88.7	75.8	63.4
University degree (%)	23.9	25.8	62.0
Average number of years in Greece	8.1	11.2	4.2

Reliability analysis (Bohmstedt, 1970; SPSS, 2007) for the 11 continuous variables of Table 2 was then used to determine the extent to which these items are related to each other, to get an overall index of the internal consistency of the scale as a whole and to identify items that had to be excluded from the scale. In fact, none of the items were excluded from the primary number of 11 items. The value of Cronbach’s alpha (α) reliability coefficient was found to be 0.87 (SPSS, 2007); thus indicating that the return migration reasons scale is reliable. Friedman two-way analysis of variance, with χ² = 2.3142 (α = 0.00) and Hotelling’s T² = 1.2968 (F = 34.26 and α = 0.00), indicated the significance in differences of item means.

Having accepted the consistency of the 11 items, the average levels of agreement for each respondent were used as the numerical values of the dependent variable ‘driver of return migration’, which, along with the categories of six independent variables, are shown in Table 4. More specifically, the dependent variable ‘drivers of return migration’ is a quantified multi-thematic variable (continuous) comprising the mean values of each of the 11 return migration reasons (ordinals). According to the Likert scale (5 = strongly agree, 4 = agree, 3 = neither agree nor disagree, 2 = disagree and 1 = strongly disagree), there are five levels of agreement for each of the return migration reasons. Thus, a high value of the dependent variable expresses a positive and strong return driver.

Table 4.

Selected independent variables.

Independent variables	Type	Categories
1. Marital status	Nominal	1 = married, 2 = not married
2. Area	Nominal	1 = Korçë, 2 = Elmpasan, 3 = Tiranë, 4 = Sarandë, 5 = Gjirokastër, 6 = Pogradec, 7 = other areas
3. Gender	Nominal	1 = male, 2 = female
4. Age	Ordinal	1 = under 25, 2 = 25–45, 3 = 45–60, 4 = over 60
5. Education	Ordinal	1 = six years or less, 2 = from seven to nine years, 3 = ten years or more
6. Years of stay in Greece	Numeric

Investigating further the decision of return migration, in order to find out how return migration is influenced by personal characteristics of the respondents, a categorical regression model was employed. It yielded R² values ranging from 0.718 (first cluster) to 0.844 (second cluster), indicating moderate relation between the ‘return migration reasons’ and the group of selected predictors. However, as R² > 0.70, it is indicated that more than 70% (from 71.8% to 84.4%) of the variance in the ‘return migration causes’, rankings are explained by the regression of the optimally transformed variables used. The F statistic values (from 8.32 to 8.46), with corresponding α = 0.00, indicates that this model is always performing well.

The relative importance measures (Pratt, 1987) of the independent variables show that the most influential factors predicting ‘return migration’ decision in the first cluster correspond to ‘education’ (accounting for 45.9%), followed by ‘area’ (23.0%) and ‘age’ (18.7%). Respectively, the relative importance measures of the independent variables, which are reported in the second cluster, are higher for the variables of ‘years of stay in Greece’, ‘education’, ‘marital status’ and ‘area’. Finally, the relative importance of the above independent variables in the third cluster is high for the variables ‘age’, ‘area’, ‘education’ and ‘years of stay in Greece’. The total percentage of the ‘return migration’ decision, which is explained by the estimated three or four independent variables in each cluster, is calculated in the last column of Table 5. In particular, the additive importance of estimated independent variables accounts for about 87.6%, 88.8% and 92.7% for the first, second and third clusters, respectively.

Table 5.

Relative importance measures.

Cluster	n	R ²	Relative importance measures				Total explanation
First	71	0.718	Education (0.459)	Area (0.230)	Age (0.187)	–	(87.6%)
Second	58	0.844	Years of stay (0.301)	Education (0.233)	Marital status (0.193)	Area (0.161)	(88.8%)
Third	71	0.763	Education (0.418)	Area (0.222)	Age (0.187)	Years of stay (0.100)	(92.7%)

Dependent variable: drivers of return migration.

Comparing results of the current study with those examined the same issue (Labrianidis and Hatziprokopiou, 2003; Labrianidis and Kazazi, 2006; Labrianidis and Lyberaki, 2004; Vadean and Piracha, 2010), it is observed that only with the most recent study of Vadean and Piracha (2010) can common results be found. A significant share of returnees (45.9%) are characterised as unsuccessful, as in our case, conforming to the failure type of Cerase (1974).

Conclusions

Globally, return migration has occupied the interest of both research experts and policy makers. In the Mediterranean Basin, where countries, especially Greece, have a short migration history as reception locations, any issue related to return migration is of special interest. Given the large-scale increase in inward migration in the last two decades and the significance of migration as a principal driver of change in southern European countries, the findings from this paper, with respect to the determinants of return migration, make an important contribution to our knowledge. However, given the difficulties of securing data, these findings with respect to different types/categories of return migrants and the factors shaping their decision to return can be seen only as indicative of the trends, and cannot be readily generalised.

In order to classify the respondents and to determine possible relations between return migration decision and several prospective reasons of return migration, an indicatory dataset has been analysed using two-step clustering and categorical regression models. The results, overall, indicate three discrete clusters of respondents with different return migration behaviour. The first cluster includes mainly ‘unsuccessful migrants’ who faced problems of racism, a lack of incorporation in Greek society and unemployment. The second cluster constitutes mainly the so-called ‘family priority immigrants or retirees’, who returned for family reasons or because of retirement. Finally, the third cluster, named as ‘younger and successful immigrants’, constitutes those who returned because they accumulated enough money for themselves or for their children’s education. The most influential factor predicting the return migration decision in each cluster corresponded to ‘education’ (10 or more years), ‘years of stay in Greece’ (seven or more) and ‘age’ (under 45 years old) for the first, second and third clusters, respectively.

In the current study, racism, difficulty of incorporation in local society, unemployment and nostalgia for home country (first cluster causes) are the most influential causes compelling immigrants to return home. It becomes obvious that return migration causes of the current study differ substantially from those found in mainstream literature on return migration. We attribute this difference to the short migration history of Greece and we presume that the same causes will be found in other countries with a short migration history.

It is interesting to identify factors that influence the decisions and status of the return migrants, which are mainly due to the short migration history of the host country. These factors stem from policy insufficiencies, such as absence of strict and unambiguous policy related to legalisation of immigrants, insufficiency of related services and lack of information about immigration issues that influence attitudes of locals against immigrants. Greek migration policies have focused mainly on controlling borders and deporting illegal migrants. The character of the policy was mainly short term without precautionary measures, a policy deficiency that made the acceptance, legalisation and social inclusion of immigrants difficult. The host society mainly confronted immigrants with racism and exclusion. Thus, we conclude that the short migration history of Greece is the factor that induced, directly and indirectly, a significant number of Albanian immigrants to return home as unsuccessful and we endorse the demands made by other scholars and organisations, concerned with migration, for an integrated migration policy that includes measures to support return migrants.

From a methodological perspective, this paper attempts to use modern multivariate methodologies to analyse return migration, including categorical techniques that are able to deal effectively with continuous and categorical variables, as well as attributes. This combination of categorical regression models with a TSCA proves very effective in analysing decision making. However, as a first systematic attempt, this study is limited to a rather small sample and time period, so generalisation is difficult. Nevertheless, the findings with respect to the significance of racism, inability to find work and more general problems associated with lack of incorporation into Greek society will, hopefully, provide stimulus for further research, especially with respect to countries that have a short migration history.

Footnotes

Acknowledgements

We gratefully acknowledge many helpful comments and suggestions from two anonymous reviewers and from the editor of the journal. The development of this paper has benefited substantially from their comments.

Funding

This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.

References

Arango

(2000) Explaining migration: a critical view. International Social Science Journal 52(165): 283–296.

Arapoglou

Sayas

(2009) New facets of urban segregation in Southern Europe: gender, migration and social class change in Athens. European Urban and Regional Studies 16(4): 345–362.

Berry

Lewis-Beck

(1986) New Tools for Social Scientists: Advances and Applications in Research Methods. Beverly Hills, CA; London and New Delhi, India: SAGE.

Bohmstedt

(1970) Reliability and validity assessment in attitude measurement. In: Summers

(ed.) Attitude Measurement. Chicago, IL: Rand-McNally & Co., pp.80–99.

Borjas

(2001) Heaven’s Door: Immigration Policy and the American Economy. Princeton, NJ: Princeton University Press.

Carletto

Davis

Stampini

. (2006) A country on the move: international migration in post-communist Albania. International Migration Review 40(4): 767–785.

Cassarino

(2004) Theorising return migration: the conceptual approach to return migrants revisited. International Journal on Multicultural Societies 6(2): 253–279.

Castles

(2004) The factors that make and unmake migration policies. International Migration Review 38(3): 852–884.

Cerase

(1974) Migration and social change: expectations and reality: a study of return migration from the United States to Italy. International Migration Review 8(2): 245–262.

10.

Charatsari

Papadaki-Klavdianou

Michailidis

(2011) Farmers as consumers of agricultural education services: willingness to pay and spend time. Journal of Agricultural Education and Extension 17(3): 253–266.

11.

Coulon de

Piracha

(2005) Self-selection and the performance of return migrants: the source country perspective. Journal of Population Economics 18(4): 779–807.

12.

Dustmann

(2003) Return migration, wage differentials, and the optimal migration duration. European Economic Review 47: 353–369.

13.

Fakiolas

(2003a) Regularizing undocumented immigrants in Greece: procedures and effects. Journal of Ethnic and Migration Studies 29: 535–561.

14.

Fakiolas

(2003b) The second legalization of the illegal immigrants in Greece. International Migration Review 37(4): 1292–1302.

15.

Fakiolas

King

(1996) Emigration, return, immigration: a review and evaluation of Greece’s postwar experience of international migration. International Journal of Population Geography 2(2): 171–190.

16.

Ghosh

(2000) Return Migration: Journey of Hope or Despair? Geneva: IOM/UN.

17.

Glytsos

(2005) Stepping from illegality to legality and advancing towards integration: the case of immigrants in Greece. International Migration Review 39(4): 819–840.

18.

Hatziprokopiou

(2004) Balkan immigrants in the Greek City of Thessaloniki: local processes of incorporation in an international perspective. European Urban and Regional Studies 11(4): 321–338.

19.

Haug

(2001) Bleiben oder Zurückkehren? Zur Messung, Erklärung und Prognose der Rückkehr von Immigranten in Deutschland. Zeitschrift für Bevölkerungswissen- schaften, Jahrgang 26(2): 231–270.

20.

Iosifides

Lavrediadou

Petrakou

. (2007) Forms of social capital and the incorporation of Albanian immigrants in Greece. Journal of Ethnic and Migration Studies 33(8): 1343–1361.

21.

Kasimis

(2008) Survival and expansion: migrants in Greek Rural Regions. Population, Space and Place 14: 511–524.

22.

King

(1986) Return migration and regional economic development: an overview. In: King

(ed.) Return Migration and Regional Economic Problems. London: Croom Helm, pp.1–37.

23.

King

(2000) Generalizations from the history of return migration. In: Ghosh

(ed.) Return Migration: Journey of Hope or Despair? Geneva: IOM/UN, pp. 7–55.

24.

King

(2004) Albania: interrelationships between population, poverty, development, internal and international migration. Mediterranee 3(4): 37–47.

25.

King

Vullnetari

(2003) Migration and development in Albania. Working Paper C5. Brighton: University of Sussex, Development Research Centre on Migration, Globalisation and Poverty. December 2003, (see http://www.migrationdrc.org/publications/working_papers.html)

26.

Kosic

Triandafyllidou

(2003) Albanian immigrants in Italy: policy implementation, coping strategies and identity issues. Journal of Ethnic and Migration Studies 29(6): 997–1014.

27.

Labrianidis

Hatziprokopiou

(2003) Return migration and economic development: the case of Albanian return migrants from Greece and Italy (in Greek). Topos 20–21: 77–97.

28.

Labrianidis

Kazazi

(2006) Albanian return-migrants from Greece and Italy: their impact upon spatial disparities within Albania. European Urban and Regional Studies 13(1): 59–74.

29.

Labrianidis

Lyberaki

(2004) Back and forth and in between: returning Albanian migrants from Greece and Italy. Journal of International Migration and Integration 5(1): 77–106.

30.

Labrianidis

Sykas

(2009) Migrants, economic mobility and socio-economic change in rural areas. European Urban and Regional Studies 16(3): 237–256.

31.

Lazaridis

Koumandraki

(2001) Deconstructing naturalism: the radicalization of ethnic minorities in Greece. In: King

(ed.) The Mediterranean Passage: Migration and New Cultural Encounters in Southern Europe. Liverpool: Liverpool University Press, pp.279–301.

32.

Lazaridis

Koumandraki

(2003) Survival ethnic entrepreneurs in Greece: a mosaic of informal and formal business activities. Sociological Research Online 8(2): 44–58.

33.

Lazaridis

Koumandraki

(2007) Albanian migration to Greece: patterns and processes of inclusion and exclusion in the labour market. European Societies 9(1): 91–111.

34.

Lazaridis

Poyago-Theotoky

(1999) Undocumented migrants in Greece: issues of regularization. International Migration 37(4): 715–740.

35.

Lazaridis

Psimmenos

(2000) Migrant flows from Albania to Greece: economic, social and spatial exclusion. In: King

Lazaridis

Tsardanidis

(eds) Eldorado or Fortress? Migration in Southern Europe. London: Macmillan Press, pp. 170–185.

36.

Mai

(2001) Italy is beautiful: the role of Italian television in the Albanian migratory flow to Italy. In: King

Wood

(eds) Media and Migration: Constructions of Mobility and Difference. London: Routledge, pp.95–109.

37.

Malhorta

(1996) Marketing Research: An Applied Orientation. Englewood Cliffs, NJ: Prentice Hall.

38.

Massey

(1990) Social structure, household strategies, and the cumulative causation of migration. Population Index 56(1): 3–26.

39.

Massey

Alarcon

Durand

. (1987) Return to Aztlan: The Social Process of International Migration from Western Mexico. Berkeley, CA and Los Angeles, CA: University of California Press.

40.

Massey

Constant

(2001) Guest workers of the 1980s and 1990s: who is going back and who is staying in Germany? In: Paper presented at the meeting of the European society of population economics, Athens, Greece, 14–17 June.

41.

Massey

Arango

Hugo

. (1993) Theories of international migration: a review and appraisal. Population and Development Review 19(3): 431–466.

42.

Michailidis

Koutsouris

Mattas

(2010) Information and communication technologies as agricultural extension tools: a survey among farmers in West Macedonia, Greece. Journal of Agricultural Education and Extension 16(3): 249–263.

43.

Moeser

(2005) Remigration von Gastarbeitern in ihre Herkunftsländer – Eine Verlaufsdatenanalyse mit dem sozio-ökonomischen Panel. Diplomarbeit, Justus-Liebig-Universität Giessen, Giessen.

44.

Nikas

Baklavas

(2009) Savings and remitting attitudes of Albanian emigrants – an exploratory data analysis. Southeast European and Black Sea Studies 9(4): 481–495.

45.

Papapanagos

Sanfey

(2001) Intention to emigrate in transition countries: the case of Albania. Journal of Population Economics 14(3): 491–504.

46.

Patiniotis

(1985) Return migration and socioeconomic integration. Greek Review of Social Research 54: 87–104.

47.

Piracha

Vadean

(2010) Return migration and occupational choice: evidence from Albania. World Development 38(8): 1141–1155.

48.

Portes

(2010) Migration and social change: some conceptual reflections. Journal of Ethnic and Migration Studies 36(10): 1537–1563.

49.

Portes

Walton

(1981) Labor, Class, and the International System. New York: Academic Press.

50.

Pratt

(1987) Dividing the indivisible: using simple symmetry to partition variance explained. In: Pukkika

Puntanen

(eds) Proceedings of the second international conference in statistics. Tampere: University of Tampere, pp.245–260.

51.

Rovolis

Tragaki

(2006) Ethnic characteristics and geographical distribution of immigrants in Greece. European Urban and Regional Studies 13(2): 99–111.

52.

Salganik

Heckathorn

(2004) Sampling and estimation in hidden populations using respondent-driven sampling. Sociological Methodology 34(1): 193–239.

53.

Sassen

(1988) The Mobility of Labor and Capital: A Study in International Investment and Labor Flow. Cambridge: Cambridge University Press.

54.

Siardos

(2002) Methods of Multivariate Analysis, 2 vols, 1st ed. Thessaloniki: ZITI Publications.

55.

SPSS (2007) SPSS 17.0 for Windows, User’s Guide. SPSS inc., Chikago, United States of America.

56.

Stark

(1991) The Migration of Labour. Oxford: Blackwell.

57.

Stark

(1997) Warum kehren migranten zurück? Mikroökonomische aspekte von rückwanderung. In: Angenendt

(ed.) Migration und flucht. Aufgaben und Strategien für Deutschland, Europa und die Internationale Gemeinschaft. Bundeszentrale fόr politische Bildung: Bonn, Germany pp.190–199.

58.

Stark

Bloom

(1985) The new economics of labor migration. The American Economic Review 75: 173–178.

59.

Todaro

(1969) A model of labor migration and urban unemployment in less developed countries. The American Economic Review 59(1): 138–148.

60.

Triandafyllidou

(2000). ‘Racists? Us? Are you joking? The discourse of social exclusion of immigrants in Greece and Italy’ in King

Lazaridis

Tsardanidis

(eds), Eldorado or Fortress? Migration in Southern Europe, London: Macmillan, pp. 186–205.

61.

Triandafyllidou

(2007) Mediterranean migrations: problems and prospects for Greece and Italy in the Twenty-first Century. Mediterranean Politics 12(1): 77–84.

62.

Triandafyllidou

(2009) Greek immigration policy at the turn of the 21st Century: Lack of political will or purposeful mismanagement? European Journal of Migration and Law 11(2): 159–178.

63.

Vadean

Piracha

(2010) Circular migration or permanent return: what determines different forms of migration. In: Epstein

Gang

(eds) Migration and Culture: Frontiers of Economics and Globalization, vol. 8. Emerald, Bingley Press, UK, pp. 497–516.

64.

Van der Kooij

Meulman

(1997) MURALS: multiple regression and optimal scaling using alternating least squares. In: Bandilla

Faulbaum

(eds) Advances in Statistical Software. Stuttgart: Lucius & Lucius, pp.99–106.

65.

Zhang

Ramakrishnon

Livny

(1996) BIRCH: an efficient data clustering method for very large databases. In: Jagadish

Mumick

(eds) Proceedings of the ACM SIGMOD Conference on Management of Data. Montreal, QC: ACM Press, pp.103–114.

66.

Zhao

(2002) Causes and consequences of return migration: recent evidence from China. Journal of Comparative Economics 30(2): 376–394.