Universiteit van Amsterdam
Immigrant
Participation on
European Labour
Markets
Assessing the determinants of Labour Market participation
Tsjangis van Oostrom
Master Thesis: Human Geography A Quantitative Approach
1
Inhoudsopgave
Immigrant Participation on European Labour Markets ... 2
Theoretical Background: Host-Society Adjustment ... 4
Neoclassical Framework ... 4
Neoclassical: Human Capital ... 5
NEM (New Economics of Migration) and Network theory ... 5
The Adaptation Process ... 6
Human Capital transferability ... 6
Contextual effects ... 7
Economic Contextual Effects ... 7
Social Contextual Effects ... 8
Policies by Host-countries ... 9
Data description and Research design ... 11
Variables and Coding ... 14
Methodology ... 19
Results and Analysis ... 19
Descriptives ... 19
Contextual effects? ... 25
Measuring the effects of Policy ... 26
Labour Market Access? ... 29
Recognition of Diploma’s and Skills & offering Study Grants ... 29
Social Security and Labour Market participation? ... 30
Conclusion ... 30
Policy ... 32
Scientific relevance? ... 33
Room for Improvement ... 34
Further Research ... 34
2
Immigrant Participation on European Labour Markets :
Assessing the determinants of labour market participation
Migration and migration policy is a fiercely debated issue at the national, as well as at the European level. Recently, a new sense of urgency has taken hold of the public as stories of refugees trying to cross the Mediterranean have become a daily occurrence. Public debates on issues related to migration have become heavily polarized and sometimes lack in empirical substantiation. This thesis sets out to fill this lacuna and to answer a broad range of questions concerning immigrant adaptation and labour market participation. Why do some immigrants find employment while others do not? Which immigrant groups have difficulty acquiring jobs and finding employment and which groups are relatively successful? Which characteristics make some countries favourable and accommodating to immigrants in comparison to other (Western) European countries? Answering these questions is of importance as Europe will remain an important destination for a broad range of immigrant groups for the foreseeable future.
Finding employment is central to successfully building a living in Europe and can serve as a stepping stone for further upward mobility and the fulfilment of other ambitions. In this thesis we will try to create a comprehensive understanding of the individual, contextual and legal-institutional
characteristics and their influence on immigrant labour market status. For this purpose we gathered a large amount of cross-country European data and designed a logistic model which includes individual as well as contextual determinants. Do certain national policies, such as diploma recognition facilities, increase the labour market participation of immigrants? Are countries with larger amounts of social spending associated with higher immigrant labour market participation? Do immigrants show higher participation with the passing of time in the host-society? These are some of the questions which this thesis aims to answer. Our model will be used to test a broad range of theories and hypothesis on human capital transferability, host-society adjustment and economic/social contextual effects. Three waves of the ESS (European Social Survey) database have been merged in order to increase the validity of our findings. For our policy assessment we will use the MIPEX1 database which has collected and assigned standardized scores on national immigrant policies across Europe, which enables us to easily compare countries. MIPEX scores allow us to understand the legal/institutional context in which immigrants operate and the labour-market policies national governments implement.
Our main research question is as follows:
How do institutional differences and policies in host countries affect the labour market participation of immigrants?
Sub-questions are:
1
3
Do certain characteristics, such as citizenship, age, educational attainment and duration of stay increase immigrant labour market participation?
Are certain contextual characteristics, like attitude vis-à-vis immigrants or social governmental spending, strong predictors of labour market participation?
Which national policies are most effective at improving the labour market position of immigrants? Which ones are not?
A large range of literature has been devoted to understanding the social and economic aspects of migrant adaptation in host-societies. We will primarily focus on economic literature and theories but some general social theories will also be used in our main framework. Hypotheses will be formulated and results will be compared with previous findings.
Why
A smooth transition onto the labour market is not only beneficial for immigrants but equally as important for host-societies. It seems likely that better economic integration reduces tensions, stimulates economic growth and improves the general attitude towards immigrants from the native population.
This thesis hopes to contribute to our understanding of European immigration in multiple ways, for example, we will try to find out which determinants are important predictors of employment status and to see, subsequently, whether we can pinpoint to specific vulnerable subgroups. Our results might also give us an indication of which host-society policies are effective in stimulating the labour market participation of immigrants. In turn, we might be able to recommend these policies to other national governments throughout Europe. Countries with a long (colonial) migration history, such as the United Kingdom and France, may offer valuable lessons to countries which have recently joined the European Union and are in the process of becoming net-migration destinations.
How
The European Social Survey (ESS) is a bi-annual survey conducted throughout Europe. It provides us with information on a large range of topics, such as labour market status, educational attainment, citizenship and duration of stay. In addition to this there are also indicators on political preferences, attitudes and well-being. Furthermore, the ESS website also offers some ‘country aggregate indicators’ such as GDP, GINI-index, HDI, etc. We will import a number of these indicators which we suspect are related to the labour market participation of immigrants.
In addition to this we will import data from MIPEX which offers country-aggregate scores on the legal-institutional framework in which immigrants operate. These are merged with our main database in order to be able to compare respondents across countries and test our hypotheses.
Descriptive and regression methods will be used in order to analyse our binary variable ‘employment status’ and its determinants. Before we start testing we will formulate a number of hypotheses to see
4
whether our assumptions and previous theories hold true for our database. The results will be interpreted and supplemented with academic literature to create a coherent narrative and to see whether we can get an overarching idea of the effectiveness of European migration policy.
Theoretical Background: Host-Society Adjustment
There are multiple phases in the migration process that all have been extensively theorized. For our theoretical overview we will draw heavily on the framework used by van Tubergen (2004) and the framework used by Portes & Rumbaut (2010). Our emphasis will be on labour market integration and in theorizing some of the determinants of labour market integration. Firstly, a short introduction of the drivers of international migration will be presented as they give a more comprehensive understanding of the migration process. This will be followed up by theories on human capital and human capital
transferability. Looking solely at human capital does not suffice (van Tubergen, 2004). In order to better understand labour market integration we will also have to touch upon the importance of contextual determinants. This will be followed up with a chapter in which we discuss and formulate hypotheses concerning the effects of European labour market policy.
Neoclassical Framework
There have been lively debates in the literature as well in the mainstream media as to what motivates migrants to make the decision to move abroad. One of the earliest attempts to conceptualize
international migration is ‘neoclassical migration theory’ (Kurekova, 2011). The underlying assumption throughout these models is that migrants are primarily driven by economic incentives and decide to migrate as they perceive the benefits to outweigh the costs. Migration flows thus are conceptualized as a global market of ‘supply and demand’. Ultimately, it is theorized, some type of equilibrium takes place where returns on migration no longer outweigh the cost of moving abroad.
Neoclassical theory is intuitively appealing but has been vigorously criticized. One of the main critiques on the ‘neoclassical model’ is that it is overly reductionist as it reduces migration incentives to economic rational decisions. Critics argue that neoclassical theory homogenizes migrants and presents migration as an ahistorical phenomenon. Furthermore, it leaves out the importance of politics and policies, which are only considered as distortion factors or additional migration costs (Kurekova, 2011). Finally, it assumes that migrants are well-informed and prepared when they move abroad while in practice migrants face an uncertain future and don’t always know where they might end up.
Neoclassical theories have been empirically tested and proven partially inadequate in explaining global migration flows (Portes & Rumbaut, 2010). Using the neoclassical framework one would assume that the poorest segments of society are most likely to decide to make the move abroad as their reward will be highest. In practice however, the poorest socioeconomic strata within sending countries cannot afford to make such a large investment, i.e. the middle or higher classes move abroad. Therefore, whenever middle-classes expand migration is likely to increase.
5 Neoclassical: Human Capital
The previously mentioned macro-level model can be transferred to the micro-level and has been termed the ‘human capital theory of migration’, early proponents of which were Sjaadstad (1962) and Todaro (1969). Human capital theory is also part of the neoclassical paradigm but incorporates
socio-demographic characteristics of the individual as important determinants of migration. The rational individual is at the centre of such analysis and moves abroad with the aim of maximizing his or her benefits and gains. Human capital such as experience, social skills and education strongly affect who makes the decision to move abroad. Different individuals in the same sending country demonstrate different propensities to migrate and have been shown to choose different receiving countries (Bonin et al., 2008). Research has shown that one of the strongest determinants of migration is education level (Bauer & Zimmermann, 1999). Migrants with high skills are more likely to decide to move abroad as they perceive their chances on the labour market as substantially larger. One of the main criticisms on human capital theory is that it fails to account for other types of migration, such as political and family
migration as it presents migration as an overly rational decision-making process. NEM (New Economics of Migration) and Network theory
The new economics of migration (NEM) presents an alternative to the neoclassical approach, with a new level of analysis and a different set of migration determinants. Instead of focusing on the individual, NEM shifted the attention to the household (Stark, 1991). NEM contends that migration decisions are not made by individual actors but typically by families or households. A broad range of conditions in the sending-country determine whether a migrant decides to depart (Kurekova, 2011). According to Stark (1991) migration must be conceptualized as a household response to both income risk and to the failures of a variety of markets, e.g. the labour or insurance market (Massey et al. 1993). Instead of looking at inequality between countries, Stark (1991) emphasized the importance of relative deprivation as a central determinant in making the decision to move abroad. Remittances are of key importance in the research done by Stark as they further highlight the concept of household interconnectedness and the ties between sending and host-countries. NEM has been criticized for focusing too much on sending-countries, being too heavily future oriented and for failing to operationalize some of the determinants of migration. Overall, the theory has not received much following or empirical testing (Kurekova, 2011). A related theory of migration, Network Theory, does not look at the determinants which initiate
migration but rather at what perpetuates migration in time and space (Massey, 1987; Massey et al., 1993). Migrant networks facilitate migration flows and can help to explain why migration continues even when wage differentials or recruitment policies cease to exist. Network theory augments earlier rational decision-making theories such as human capital theory. Diaspora and ethnic networks are being
highlighted as important determinants of destination choice (Massey et al., 1993). The network theory also helps to explain why migration patterns are not evenly distributed across countries, but rather how they form geographical clusters (Faist, 2000). Transnational social networks reduce costs for potential migrants as successive waves of immigrants make adapting and adjusting into host-societies easier. Transnational ties are being utilized, information can be exchanged and financial support can be offered. Faist (2000) emphasized the cultural practices of diaspora and ethnic networks to maintain ties to their homeland by using symbols and rituals that tie them to their homeland. These transnational networks
6
explain why immigrants often originate in specific sending countries and end up in specific destination-countries.
The Adaptation Process
Human Capital transferability
When immigrants arrive in a country they may find that the human capital they brought with them is not immediately useful and cannot be easily transferred across countries. With the passing of time human capital previously acquired in the country of origin can be more effectively transferred to the host-society, thereby increasing the economic chances of migrants. As migrants get more familiar with their surroundings, expand their social network and acquire language skills they become more attractive for employers. The process whereby migrants slowly adjust to the host-society and increase their economic participation with duration of stay has been called the ‘Dip and Catch-up model’. This model has been tested in numerous studies, e.g. (Chiswick, 1978; Borjas, 1995; Lundborg, 2013; Zorlu, 2013). In the US employment gaps (native-immigrant) diminish quickly but significant wage gaps persists, while in the European mainland it is exactly the opposite, employment gaps persist while wages differentials are smaller (Lundborg, 2013).
The human capital immigrants bring along determines the relative success of their economic incorporation into the host-society. The importance of certain factors such as previous experience, education, language skill and duration of stay has been well documented and empirically tested, e.g. (Chiswick et al., 1997; Chiswick & Miller, 2003) and van Tubergen (2004). In what follows we will shortly discuss the importance of individual characteristics, such as education, language, citizenship and gender.
Education has been found to be one of the strongest determinants of immigrant (and native) labour market participation, e.g. (van Tubergen, 2004). A relatively new line of research has sought to make a distinction between education acquired before and after arrival, as there are indications that education acquired after arrival is more rewarding in terms of earnings (Bratsberg & Ragan, 2002; Friedberg, 2000; Basilio, 2010). Unfortunately, our ESS database does not allow for differentiation between education acquired before and after arrival. Nevertheless, we expect to find a strong positive relation between education and labour market status.
Kossoudji (1988) has investigated the importance of language proficiency on the American labour market. She found that there is an earnings penalty for immigrants who do not acquire the host-society language, but this penalty differs among ethnicities and occupations. Immigrants operating within an ethnic niche experience smaller penalties on language deficiency. Chiswick and Miller (2002) found similar effects, looking at linguistic concentrations and the effects on earnings. English language proficiency significantly increased earnings, all other things being equal. The return to other human capital was also found to be higher for those respondents with English language proficiency, suggesting some kind of complementarity effect. Furthermore, respondents living in areas that were linguistic concentration zones showed lower earnings, highlighting the importance of residential choice.
7
These findings seem to point to the fact that without a strong command of the spoken and written language immigrants are limited in their business network, have limited knowledge about job opportunities and are more restricted in the jobs they can access.
The importance of citizenship is an issue that is touched upon in earlier work done by Chiswick (1978). Chiswick found a significant negative effect (15%) on earnings for immigrants who did not hold American citizenship. It is theorized that immigrants who acquire citizenship are more committed and have the security and ability to make long-term plans for improving their position on the labour market. In addition to this, citizenship might make you eligible for government assistance in education, language acquisition or social security. Hence, we expect citizenship not only to impact earnings but also to influence the ability to find employment, thus we hypothesize that immigrants without host-country citizenship are less likely to be employed.
The importance of gender on the labour market participation of immigrants is a central theme in the work done by Antecol (2000). Antecol examined the gender-gaps in labour market participation in the US. She was interested in whether these differences can be attributed to differences in human capital or whether they were attributable to preferences in family structure and (traditional) gender roles. Antecol finds evidence that the gender-gap is primarily attributable to preferences and not to human capital. Larger gender-gaps in countries of origin are associated with larger ‘gender-gaps’ in host-societies. Migrants originating in countries with large gender gaps, such as Afghanistan or Jordan often re-establish these patterns in the host-society. However, she also found evidence that these gender-gaps diminished among the second (or higher) generation. Continuing in this line of reasoning, as our research focuses on first generation immigrants, we expect to find substantial gender gaps.
Contextual effects
Human capital can only partially predict labour market status, as various ethnic groups in different countries all perform differently on the labour market (van Tubergen, 2004). This seems to suggest that in order to better understand economic incorporation we have to acknowledge the importance of contextual factors as well. In the following section we will shortly discuss two contextual dimensions which we hypothesize impact the economic participation of immigrants. We made a distinction between economic and social contextual effects.
Economic Contextual Effects
Pedace et al. (2012) hypothesized that there is a ‘scarring effect’ when migrants arrive during periods of economic downturns. It is theorized that migrants who arrive during economic contractions experience difficulties finding a job and in turn have little opportunity of gaining experience and thus making headway on the labour market. The empirical findings in his research show that there is indeed a difference in propensity to find employment amongst different arrival cohorts. However, these employment gaps diminished as the duration of stay increased, essentially approximating or even exceeding native levels of employment (in the US) after more than ten years of residence. Similar results
8
were found in the work done by Chiswick & Miller (2002) confirming an initial scarring effect that diminishes with a longer duration of stay.
Another economic contextual effect we will incorporate in our research is the size of government expenditures. Is the size of government expenditures in some way related to the labour market status of immigrants? In order to further illustrate this we will use the framework originally proposed by Esping-Anderson (1990). Esping-Anderson has created a typology in which he presented 3 types of states: the Liberal state, Corporatist-states and Social-Democratic states. Liberal states are characterized by flexible labour jurisdiction, limited unionism and modest welfare assistance where entitlement rules are usually strict. Corporatist states usually offer welfare and assistance based on previously made contributions and depending on which labour association you are affiliated with. The corporate group is regarded as the linchpin of society and the state is regarded primarily as intermediator. Welfare states are most
generous, providing relatively high levels of welfare and service, high levels of unionism and having rigid labour market jurisdiction protecting employees.
We hypothesize that states which offer assistance after arrival, be that financial or educational, encourage the economic incorporation of immigrants in the labour market. Therefore, we suspect that immigrants living in social-democratic welfare states would thus be more likely to be employed when compared to immigrants living in liberal or corporatist states. We hypothesize government expenditures trickle down to immigrants which enable them to quickly adapt to local circumstances and actively search for jobs.
However, it might also be theorized that immigrants perform better in countries with flexible regulations and limited unionism as it enables immigrants to easily set-up their own businesses and compete with natives. In liberal states it might be expected from immigrants that they quickly become self-sufficient, as social security is only marginally available. We will touch upon this issue later when we discuss national integration policy, for now it suffices to say that we suspect immigrants to fare better in social-democratic states, i.e. states with relatively large social expenditures.
Social Contextual Effects
Even though there has been some research on discrimination and the position of immigrants in the labour market (e.g. Model & Lapido, 1996; Wilson & Martin, 1982), there is no single dominant discrimination theory (van Tubergen, 2004). Some theories are concerned with the role of the state while others emphasize the role of individuals. The basic assumption in these theoretical frameworks is that they look at in-group preferences and out-group prejudices. It is theorized these out-group
prejudices restrict immigrant participation as jobs or opportunities are being offered to members of the in-group (van Tubergen, 2004). These prejudices might also impact earnings, for example when delaying promotion or restricting access to higher occupational strata.
In an article titled ‘Does attitude towards immigrants matter?’ Waisman and Larsen (2008) found that negative attitudes towards immigrants impacted immigrant earnings. The reasoning was as follows: immigrants preferred to live among co-ethnics in neighbourhoods where they could escape negative prejudices and sentiments. This in turn had ramifications for their occupational choice and career
9
progression, which in turn impacted their earnings (on average 12%). Employers might decide not to hire certain potential employees based on prejudices or perceived differences. Employees, in turn, might prefer not to work for certain employers as they expect confrontations or negative attitudes in the workplace. We have incorporated a measurement tool for native attitudes towards immigrants in our analysis. We suspect that positive attitudes are associated with increased labour market
opportunities and higher participation of immigrants.
Policies by Host-Countries
One of the relatively uncharted domains in the academic literature is the legal/institutional framework in which migrants operate upon arrival and the effects on labour market participation. MIPEX, the migrant integration policy index,2 provides us with an opportunity to use standardized measures in order to compare European countries on immigrant labour market policies. These standardized scores have been assigned after consultation with legal experts in each of the respective countries and are updated on a yearly basis. In our research set-up we have chosen to focus on five policy indicators which we think affect immigrant labour market participation.
In the following section we will shortly discuss each policy indicator and formulate hypotheses regarding their effect on the economic incorporation of immigrants. We will present an overview table with country-scores of our policy indicators in the next chapter.
1. Immediate equal labour market access for Labour and Family migrants.
When immigrants arrive in a new host-country, they often have to go through an initial waiting period before they are allowed to participate and contribute. This initial waiting period can range from a couple of months to multiple years depending on which country an immigrant arrives in.3 We suspect that granting immediate access to the labour market can significantly improve the socioeconomic position of immigrants. Immigrants who are not allowed to participate remain largely unproductive until the waiting period is over, after which they might have lost important economic opportunities.
MIPEX makes a distinction between ‘Labour migrants’ and ‘Family migrants’. Labour migrants often face an initial waiting period while the waiting period for Family migrants is shorter. Legislature differs across countries, e.g. full labour market access upon arrival for both family and labour migrants is found in: Finland, Portugal and Spain.
2. Equal access to public sector jobs.
A sizeable share of European countries prohibit non- (EU) citizens from working in the public sector. Examples of public sector employment are jobs in education, administration, the health sector or governmental agencies. Public sector employment constitutes a considerable share of the domestic
2 MIPEX: Migration Integration Policy Index. Available at: http://www.mipex.eu, last accessed: 6-8-15’ 3
MIPEX: Migration Integration Policy Index. Available at: http://www.mipex.eu/labour-market-mobility last accessed: 6-8-15’
10
economy, Hammouya (1999) estimates that 22% of employment in developed countries is public sector employment, with rates even higher (30%-40%) for countries in developmental transition. Hence, we hypothesize that countries which restrict access show lower rates of immigrant employment. Only after naturalization are immigrants in these countries allowed to apply for public sector jobs.
Public authority functions, such as police officers, employees in the armed forces or judges are per definition non-accessible for non-citizens in most countries, this is taken into account when determining country-scores on this indicator.
Examples of countries which completely bar access for non- (EU) citizens to public sector jobs are: France, Slovenia and Poland. Countries which do grant full access are: the United Kingdom, the Netherlands and Denmark.
3. Equal access to study grants for all.
It seems reasonable to assume that granting study grants can stimulate upward mobilization for
immigrants as this enables them to pursue education that would otherwise be too expensive. However, education requires significant time-investments which in turn might reduce their (immediate) activity on the labour market. Therefore, we hypothesize that countries which offer study grants are associated with lower labour market participation but might have higher rates of immigrant participation in higher occupational strata. Unfortunately, we do not have an indicator of occupational status but we will look at employment status and see whether ‘offering study grants’ is associated with lower participation. Examples of countries which offer study grants to non-citizens are: Sweden, Portugal and Norway.
4. Facilitated recognition of qualification and skills.
Throughout the 20th century there has been a large expansion of tertiary education worldwide, with developing countries rapidly extending their universities and improving their curricula (Freeman, 2009). This has resulted in a large group of immigrants with tertiary degrees which are not always
acknowledged or recognized in European countries. We suspect that countries which do not have official recognition procedures impediment immigrant labour market access. Therefore, we hypothesize that these countries are likely to show lower participation rates.
Procedures to recognize skills and foreign degrees are relatively new and only facilitated in traditional countries of immigration such as Germany, the United Kingdom and Sweden. In the Netherlands recognition of diplomas is facilitated by the NUFFIC4. Programs to mainstream these efforts across Europe have been initiated by the European Commission5, but a majority of countries in the EU do not yet, or only partially, facilitate the recognition of foreign degrees and skills.
5. Equal access to social security.
4
NUFFIC: Netherlands university foundation for international cooperation. Accessed at: 29 May, Available at: https://www.nuffic.nl/en/expertise/mobility-statistics
5
Directive 2005/36/EC of the European Parliament and of the Council of 7 September 2005 on the recognition of professional qualifications, Available at: http://eur-lex.europa.eu/legal-content/EN/TXT/?uri=celex:32005L0036
11
In some European countries non-citizens have the opportunity to claim social security, such as:
unemployment benefits, old age pensions, invalidity benefits, maternity leave, family benefits and social assistance. One could argue that this might enable immigrants to build up a stable and secure living and to search for jobs which correspond to their talents and motivations. Others might argue that providing social security might impediment labour market participation as it reduces incentives and urgency. Therefore, it is difficult to forecast what the effects of offering social security on immigrant labour market participation might be. We suspect to find a neutral or slightly negative association between offering social security and immigrant labour market participation.
Countries that do offer full social security for immigrants are: Norway, Spain and Sweden. Most
countries do offer some limited form of social security as is the case in for example: the United Kingdom, Slovakia and Belgium.
-
In order to further illustrate the effects of immigrant labour market policy MIPEX has created a ‘best case scenario’ and a ‘worst case scenario’:
Best Case:
“A migrant with the right to work and live in the country has the same chances as everyone else in the labour market. From day one in the country, she and her family members can start applying for any job in the private or public sector. She gets her qualifications from abroad recognised. She can then improve
her skills through training and study grants. The state encourages her by targeting her specific needs - for example, she can take language courses focused on her profession. Job mentors and trained staff help her assess skills and use public employment services. Once employed, she has the same rights as all
workers in the country.”6
Worst Case:
“Where a migrant cannot fully contribute to the country’s economic life, his skills and ambitions go to waste. He must wait 5 years to have the same right as nationals to work, study or start his own business.
Even then, he is barred from working in many sectors and professions. In the meantime, he has to look for work on his own, without any general or targeted support. Because his foreign qualifications are not
recognised, he may have to give up his career to take whatever job he finds. Employers do not have to provide him with the same working conditions or social security as his co-workers.”7
Data description and Research design
6 MIPEX: Migration Integration Policy Index. Last accessed: 28 May, Available at:
http://www.mipex.eu/labour-market-mobility last accessed: 5-8-15’
7
12
Our research will make use of the European Social Survey which is a bi-annual questionnaire covering a broad range of topics. The survey has information regarding basic demographics but also provides us with information on a wide range of political and sociocultural attitudes. Some of the demographic indicators we utilize in our analysis are age, gender, duration of stay and educational attainment. These indicators will all be regressed on our main dependent variable ‘labour market status’.
As we are primarily interested in immigrants it was deemed useful to merge multiple ESS waves (08’ – 10’ – 12’). We did this in order to increase the number of immigrants in our database which may add to the validity of our findings. Cross-country comparisons can be made and new indicators can be deduced from the available responses in the ESS database. The ESS webpage has recently also started to offer some country-aggregate indicators which will be utilized in our model.8
In order to be useful for our research countries had to meet a number of requirements. First of all they had to participate in each of the three ESS waves (08’ – 10’ – 12’), otherwise we might get biased conjuncture effects. Secondly, as we are primarily interested in immigrants, we excluded countries which barely had any immigrants in our database, e.g. Albania. Finally, we excluded some post-communist states which hosted significant amount of ‘immigrants’ which were in fact ethnic Russians, Belarusians or Ukrainians who decided to stay post break-up and were suddenly considered
‘immigrants’ by their new respective host-countries.
Shown below is a map of countries that we decided to include in our research:
Map 1. Countries meeting all three ESS requirements
8
13
We would have preferred to include Italy, Greece, Austria and Iceland but these countries failed to meet our criterion of full participation in all ESS rounds. We have decided to include Slovenia because in our sample Slovenia hosts quite a large amount of post-civil war Yugoslavian immigrants, e.g. Bosnians, Croatians and Serbs. In addition to this, relative favourable economic conditions turned Slovenia into a destination country for immigrant populations from Macedonia, Albania and Italy.
We felt Czech, Slovakia, Poland and Hungary could remain, because judging from our sample they host significant amounts of recent immigrants from Eastern Europe, such as Romanians, Bulgarians and Ukrainians. We think including these countries is justifiable because these countries are in a transition phase of becoming net-immigrant destination countries (Drbohlav et al., 2014). Recently, the European Commission ordered EU member-states to host refugee populations according to new pre-determined quota. The national parliament of Hungary has adamantly resisted such plans and has decided to legislate for an opt-out clause claiming that Hungary’s asylum system is already overburdened.9 Events
9
Reuters, UK. “Defying EU, Hungary suspends rules on asylum seekers”, available at:
14
such as these exemplify the tensions that exist between integration efforts by the European Commission and national governments that are faced with increasing numbers of arrivals. This is why we consider it interesting and justifiable to include these (new) accession states in our analysis.
Variables and Coding
Individual demographic characteristics can all be found inside the ESS database. In some cases we have reduced the number of categories in order to be able to present clearer results, as is the case with education, religion and years since migration. We transformed some continuous variables into
categorical variables which makes understanding the results and the impact on our dependent variable more easily interpretable.
Immigrants are such a heterogeneous population that we have decided to create four subpopulations: ‘Western’ immigrants, ‘Mediterranean’ immigrants, ‘Refugees’ and ‘Brain-Drain’ immigrants.
15
For our Western subpopulation we included migrants originating in the US, Canada, Australia, South Korea, Singapore, Japan and New Zealand. We decided to exclude intra-European migrants because we suspected the socio-demographic profile of intra-European migrants to be more heterogeneous compared to immigrants originating from overseas.
Our second subpopulation is immigrants from Mediterranean countries, which are frequently described as labour migrants in the European context. Countries in this subsample include: Morocco, Algeria, Libya, Egypt, Lebanon, Turkey and Tunisia.
Our ‘refugee’ population includes immigrants from countries which have suffered from civil unrest/wars and or political suppression, we included: Afghanistan, Iraq, Iran, Eritrea, Sudan, Somalia, D.R. Congo, Rwanda, Angola, Sri Lanka, Ethiopia and the Central African Republic. Others could be included but we decided this would suffice.
Finally, we have created one category for countries commonly associated with Brain-Drain. We feel this is an interesting group, sometimes overshadowed in academic literature by other migration flows. We have chosen our countries based on an OECD publication10 and on personal judgement. We decided to include: India, Philippines, China, Israel, South Africa, Malaysia, Hong Kong, Taiwan and Indonesia. We also included Indonesia, even though we suspect human capital from Indonesia to be more
heterogeneous. We could have included others such as Russia, Bangladesh and/or Nigeria but felt this would diminish the strength of our findings due to heterogeneity.
NB. Not all immigrants hailing from ‘refugee’ sending countries are refugees. Some of them might be exchange students deciding to prolong their visit, family reunification migrants or professionals. This typology only serves as a handy method to extrapolate and magnify our results, i.e. to see whether we can create a useful distinction between migration flows and origin groups.
Shown below is a table in which all the variables we will use in our analysis are presented, followed by the variable names used in the ESS dataset and our coding.11
Table 1. List of variables used in our models, includes main variable, individual/contextual and policy variables. Non-relevant categories have been excluded in this table.
10 Dumont, J.C. (2011) World migration in Figures. OECD-UNDESA. Available at:
http://www.oecd.org/els/mig/World-Migration-in-Figures.pdf
11
16 Dependent variable:
Employment status can be found under pdwrk in the dataset. This variable looks at the main activity of respondents during the last seven days. Our main dependent variable pdwrk does not differentiate between active and inactive populations, thus it includes respondents whose main activity is education or taking care of the household.
Therefore, it was deemed useful to have a control variable which is limited to the active population. This variable can be deduced from mnactic which also focused on the main activity of respondents during the last seven days. This variable excludes those working in the household, doing community service, following education or military draft. In both cases we imposed an age restriction of 15 to 65. A distribution of mean values of our control variable across countries can be found in the appendix (Nr.1). When looking at the appendix, Switzerland, Norway and the Netherlands show high percentages (97%) of employment while Ireland (81%), Portugal (83%) and Spain (86%) show (much) lower levels. Observed immigrant employment levels are on average 5% lower. France, Spain and Denmark show large native-immigrant employment gaps of around 9%.
Variables Dataset Coding
Employment status pdwrk 1: Employed 0: Unemployed (Total population, age 15-65)
work 1: Employed 0: Unemployed (Active population, age 15-65)
Education edu 1: Primary 2: Lower Secondary 3: Higher Secondary 4: Vocational 5: Tertiary Cohabitation cohabit 1: Lives Together with Partner 0: Lives alone
Language language 1: Speaks official language at home 0: Speaks non-official language at home Years Since Migration ysm_cat 1: 0-5 Years 2: 6-10 Years 3: 11-20 Years 4: 20+ Years
Religion religion 1: Catholic 2: Protestant 3: Eastern Orthodox 4: Islamic 5: Non-Religious Gender gndr 1: Male 2: Female
Citizenship ctzcntr 1: Citizen 2: Non-Citizen Born in Country brncntr 1: Yes 2: No
Origins origins 1: Western 2: Mediterranean 3: Refugees 4: Brain-Drain Essround essround 4 5 6
Gross Domestic Product gdp Nominal GDP per capita, in US $ GDP Growth gdp_growth Average annual Growth rates 08' - 12' Social Expenditures as % of GDP social_exp Scale 1-100 (1= None 100=All) Attitude Towards Migrants attitude Scale 1-10 (1= Hostile 10= Welcoming) Policy variables:
Immediate Access to Lab. Market immediate_access 1: Neither 2: Family 3: Family and Labour Migrants Access to Public Sector access_public 1: None 2: Partial 3: Full Access
Access to Study Grants access_studygrants 1: None 2: Partial 3: Yes Recognition of Foreign Diplomas recognition_diplos 1: Weak 2: Partial 3: Strong Access to Social Security access_security 1: Weak 2: partial 3: Full
17 Individual variables:
Education is based on ISCED scores, which is a standardized indicator of educational attainment. ISCED scores make comparison between European countries possible. We have reduced the number of categories in order to present clearer results.
In the ESS database there is no direct indicator of host-country language-fluency so we tried to create a proxy-measurement by looking at the language that is spoken at home. Respondents who speak the official language at home are coded as (1) and those who do not are coded as (0). This is an imperfect measurement of ability but the only approximation the ESS database offers for language fluency. We have reduced the number of categories in our religion variable to only include those categories that we deemed insightful. The number of respondents that followed Eastern religions/Judaism/other Christian denominations were too small to draw meaningful conclusions from, which is why we decided to exclude them from our analysis.
Contextual variables:
We assigned country-level scores to individual respondents. For example, all respondents in Germany are assigned a GDP-score of: 41547.
For our economic indicators GDP and Social Spending as % of GDP we consulted the online ESS library.12 GDP growth figures are based on UN data.13 Country scores can be found in the appendix (No. 2). We created one ‘social indicator’ which we termed Attitude. Attitude is a composite variable measuring attitude towards immigrants. For this we used two available variables in the ESS dataset:
Variable imwbcnt: “Immigrants make country worse or better place to live in” Variable imueclt: “Country’s cultural life undermined or enriched by immigrants”.
Theoretical scores ranged from 1 to 10. Higher scores indicate positive attitudes while lower scores indicated negative attitudes towards immigrants. Mean scores were calculated by country. Sweden, Finland and Poland score highest while Hungary and Czech score lowest (Appendix No. 2).
Policy variables:
MIPEX provides us with standardized scores on the previously presented policy domains. We have chosen five policy indicators which we hypothesize will influence the employment status of immigrants. Shown below is a table which depicts country scores on our five policy-indicators.
12 Country-level GDP & Social Spending can be found here: http://www.europeansocialsurvey.org/data/multilevel/ 13
Country level GDP growth figures, available at:
18
Table 2. MIPEX overview table of country legislation on 5 policy indicators.
We recoded scores from our MIPEX overview table in order to be able to include them in our analysis. Scores range from 1 to 3, where 1 is considered unfavourable and 3 is considered favourable. In order to further illustrate this, Finland is assigned a score of (3) on our first policy indicator (Immediate equal labour market access for Family and Labour migrants), whereas Ireland is assigned a score of (1). Exact coding can be found in the appendix (No.3).
As can be seen from the overview table, new EU accession countries bar access to the labour market for new arrivals, such as for example Slovenia, Slovakia and Hungary. Most countries have some
intermediate form in which they allow family migrants but delay labour market access for (temporary) labour migrants. Delaying access to the labour market can range anywhere from a couple of months to several years.14
Access to working in the Public Sector is prohibited for those without EU-citizenship in Poland, France, Slovenia and Slovakia. Partial access is granted for non-EU citizens in Germany, Ireland and Swiss. Full access to the public sector is granted in Scandinavian countries, the UK and Switzerland.
Some countries do not yet have official and professional diploma and skill recognition procedures, for example Switzerland and France.15 We suspect this to significantly delay access and impediment participation on the labour market, which is something we will investigate in the coming chapters.
14
MIPEX labour market mobility: available at: http://www.mipex.eu/labour-market-mobility
15
See: http://www.mipex.eu/switzerland & http://www.mipex.eu/france
Immedi a te equa l l a bour ma rket a cces s for a l l tempora ry l a bour & fa mi l y mi gra nts *
Equa l a cces s to publ i c s ector
Equa l a cces s to s tudy gra nts for a l l
Fa ci l i a ted recogni tion of qua l i fi ca tions a nd s ki l l s (s core)
Equa l a cces s to s oci a l s ecuri ty for a l l
Belgium Fa mi l y Pa rtia l Pa rtia l Pa rtia l Wea k
Czech Fa mi l y Yes Pa rtia l Pa rtia l Wea k
Denmark Fa mi l y Yes Pa rtia l Pa rtia l Pa rtia l
Finland Both Yes Yes Pa rtia l Pa rtia l
France Fa mi l y None Pa rtia l Wea k Ful l
Germany Fa mi l y Pa rtia l Pa rtia l Strong Ful l
Hungary Nei ther Pa rtia l None Wea k Pa rtia l
Ireland Nei ther Pa rtia l None Wea k Pa rtia l
Netherlands Fa mi l y Yes Pa rtia l Strong Ful l
Norway Fa mi l y Yes Yes Pa rtia l Ful l
Poland Fa mi l y None Pa rtia l Wea k Wea k
Portugal Both Yes Yes Pa rtia l Ful l
Slovenia Nei ther None None Pa rtia l Wea k
Slovakia Nei ther None Yes Wea k Wea k
Spain Both Yes Yes Pa rtia l Ful l
Sweden Fa mi l y Yes Yes Strong Ful l
Switzerland Fa mi l y Pa rtia l None Pa rtia l Ful l
19 Methodology:
Our analysis is very similar to earlier work done by Chiswick (1997) in which he investigates employment status and duration of stay. Our dependent variable is dichotomous, i.e. (1) employed (0) unemployed, therefore we decided to use binomial logistic regression techniques. We created multiple models predicting the chance of being employed, in order to do this we have included a wide range of
individual/contextual and policy determinants. Observed (logit) coefficients give us a sense of strength and influence on the likelihood of being employed. We estimated separate models for our different subpopulations and follow this up by separate models for policy domains.
First, descriptive tables will be presented and interpreted to get an impression of the differentiation and different socio-demographic characteristics of our subpopulations. This will be followed up by logistic models to assess the influence individual and contextual determinants have on the chance of being employed. The second section of our analysis will be concerned with assessing the influence of host-country policy on immigrant chances of being employed.
Results and Analysis
Descriptives
For our initial assessment we present tables on the characteristics of our native as well as for our four subpopulations and their mean observed values on individual variables.
20
Table 3. Descriptive table on individual variables, shows proportions of population on selected indicator, by origin groups.
Employment ranges widely across groups. Immigrants originating in Western countries (e.g. Australia, US, Canada) show high rates of employment (97%) while immigrants from countries below the Mediterranean and Refugee-sending countries, show much lower employment rates (81%-86%). Education also varies substantially, more than half of Western immigrants have tertiary degrees, while only 12% of our Mediterranean sample has completed tertiary education.
Natives Western Mediterranean Refugees BrainDrain
Employed (Active Pop.) 0.92 0.97 0.81 0.86 0.96
Education: Primary 0.13 0.02 0.29 0.15 0.14 Lower Secondary 0.19 0.06 0.21 0.21 0.17 Higher Secondary 0.38 0.19 0.31 0.28 0.18 Vocational 0.11 0.22 0.07 0.06 0.16 Tertiary 0.18 0.51 0.12 0.30 0.35 Age (Mean)* 41 41 41 37 40 Language** 0.97 0.87 0.55 0.48 0.45
Years Since Migration
0-5 Years 0.18 0.09 0.16 0.23 6-10 Years 0.09 0.14 0.15 0.16 10-20 Years 0.17 0.18 0.36 0.18 20> Years 0.56 0.60 0.33 0.43 Male 0.48 0.39 0.53 0.48 0.49 Citizenship 0.99 0.53 0.59 0.66 0.61 Married/Cohabiting 0.62 0.68 0.70 0.52 0.66 Religion Catholic 0.39 0.15 0.10 0.18 0.13 Protestant 0.16 0.18 0.01 0.12 0.10 Eastern-Orthodox 0.00 0.00 0.01 0.03 0.00 Islamic 0.01 0.00 0.62 0.30 0.08 Non-Religious 0.42 0.55 0.23 0.30 0.31 Other 0.02 0.11 0.03 0.08 0.38 N 98594 273 768 442 410 * Excluding retirees
21
Immigrants originating in what we termed BrainDrain countries have quite high rates of tertiary degrees. On average, BrainDrain immigrants are relatively recent arrivals while a majority of the Mediterranean population has been in Europe for over 20 years.
Our refugee population is relatively young, on average 4 years younger than our other subpopulations. Concerning gender, Western immigrants are much more likely to be female (61% of our sample), while other subgroups show a more even distribution. Cohabitation is lowest among refugees and highest among our Mediterranean sample.
Citizenship rates are lowest amongst Western immigrants which might be explained by the fact that Western immigrants already hold citizenship of another Western country which enables them to travel and cross borders relatively easily.
Religious characteristics of immigrant groups differ widely. Immigrants from Western countries are predominantly non-religious, while immigrants originating from below the Mediterranean are predominantly Muslim. Contrary to our expectations, a large group of immigrants from our refugee subpopulation is non-religious. Only a small percentage of our total sample is Eastern-Orthodox, which might be explained by the fact that we excluded a majority of post-communist Eastern European states. For a more thorough analysis we are going to apply logistic analysis. We have created a comprehensive model which includes individual and contextual predictors on our main dependent and control
variable.16 Our main models will be presented below and the ones using the control variable can be found in the appendix (No.5 & No.6). We have added an Essround variable in order to account for conjuncture effects.
The results of our Logit analysis on our main dependent variable (total population) are as follows:
16
22
Table 4. Logit analysis using our dependent variable (pdwrk), including individual and contextual determinants, separate models by origin groups.
Natives Western Mediterranean Refugees Brain-Drain
Education: Primary (Ref) (Ref) (Ref) (Ref) (Ref)
Lower secondary 0.27*** -1.76 0.43 -0.09 1.25 Upper secondary 0.85*** 0.14 0.55 1.52** 1.94** Vocational 1.15*** 1.35 1.53** 0.51 1.34 Tertiary 1.59*** 1.03 0.92* 1.86** 1.93* Gender: Female -0.61*** -2.02*** -0.88*** -1.21** -0.74 Citizenship: No 0.04 -0.85 0.01 1.76*** -0.20 Language 0.10 -0.10 0.09 0.47 -0.07
Years Since Migration
0-5 Years (Ref) (Ref) (Ref) (Ref)
6-10 Years -0.70 0.07 0.70 0.48
11-20 Years -0.08 0.09 2.42*** 0.49
More than 20 Years 0.78 0.28 2.89*** 0.37
Age 0.43*** 0.24 0.34*** 0.24* 0.51***
Age^2 -0.01*** -0.00 -0.00*** -0.00** -0.01***
Cohabiting/Marriage 0.31*** -1.21 0.18 0.52 0.06
Religion: Catholic (Ref) (Ref) (Ref) (Ref) (Ref)
Protestant 0.30*** 2.62** -1.23 -1.15 -1.03
Eastern Orthodox -0.17 (empty) -1.05 -2.69** (empty)
Islamic -0.48** (empty) -1.03* -2.52*** -2.10* Non-Religious 0.08** 1.97** -0.34 -0.27 -1.20 Essround: 4 Round 5 -0.10** -0.19 -0.15 -0.43 -0.80 Round 6 -0.08* -0.23 -0.17 -1.01* -0.88 Contextual-level: GDP 0.00*** -0.00 0.00 0.00 -0.00 GDP Growth 0.05*** 0.59 -0.37 -0.05 0.22 Social Expenditures as % GDP 0.01 -0.59 -0.29 0.67* 0.08 Attitude -0.00 -0.25** -0.05 -0.09 0.07 Constant -8.17*** 6.42 -3.72 -7.08* -8.59** N 74962 241 691 404 362 Pseudo r^2 0.23 0.18 0.14 0.28 0.18 legend: * p<.05; ** p<.01; *** p<.001
23
As can be seen, education is a strong predictor for labour market status. This finding seems to reaffirm earlier research into the relationship between education and labour market status, e.g. Chiswick et al. (1997) and Chiswick & Miller (2003). This finding does not only hold true for natives but is equally observed for our immigrant subpopulations.
There is quite a strong and significant penalty for females across all groups. The lower observed rates are partially explained by our dependent variable which includes inactive groups such as those who work in the household. When looking at our control model, the active population, these results are much less pronounced but nevertheless still remain. Our findings seem to confirm the earlier work done by Antecol (2000) which described lower participation of females.
We hypothesized that there is a penalty for immigrants who do not acquire citizenship, as earlier discussed in the works of Chiswick (1978) and Chiswick & Miller (2002). Our findings are unclear and do not seem to confirm this. Our refugee population shows a significant positive effect for those
immigrants who do not (yet) hold citizenship, meaning that the chances of being employed as an immigrant hailing from a refugee sending country are higher when they do not hold citizenship. This is a confusing result. We speculate that this might be explained by the fact that refugees who acquire citizenship proceed to invest in education. Or alternatively, the easier employability of refugees without citizenship which can reduce costs for employers, for example in agriculture or manufacturing.
Language does show some positive coefficients, albeit non-significant. Especially for refugees, the language spoken at home does influence labour market status. Refugees that speak the official language at home have a slightly higher chance of being employed. Further research into this topic might be useful.
Our predictor ‘Years Since Migration’ shows strong effects on labour market status for our refugee subpopulation, which seems to confirm the earlier mentioned Dip and catch-up model. With the passing of time migrants acquire language proficiency, knowledge of the labour market and affinity with the host-country, which increases their likelihood to find employment (e.g. Chiswick, 1978; Uhlendorff and Zimmermann, 2006).
There seems to a strong bonus for people who are cohabiting. There are a number of possible
explanations for this, one is that cohabitants might be more determined to find and hold on to jobs, as they often have children or other responsibilities. Cohabitation can offer some continuity and stability in life which does seem to have a benevolent impact on employment status.
Age in general seems to be positively associated with labour market status. In order to assess the strength of this variable on employment status we have created a plot which illustrates the observed effects. We included a squared age variable which accounts for diminished participation as respondents near retirement age. NB. Y-axis has been shortened to magnify observed effects.
24
Illustration 1. Predicted employment status for male native university graduates accounting for age and different religions.
As can be seen, there is a sizeable gap between Muslims and other religious denominations. This is a recurring find amongst natives as well as all immigrant groups. A large range of literature has been devoted to understanding this particular find (e.g. Alba et al., 2008; Cheung, 2014). Frequently, authors point to discrimination on the labour market, (lower) social-class origins, less information on
opportunities, smaller business networks or difficulty signalling human capital as possible explanations. Our results seem to suggest that irrespective of human capital such as education or language ability, there still seems to a sizeable gap solely attributable to religion.
Qualitative work done on Muslim immigrants and the descendants of immigrants might offer some insights here as to how this process plays out. Ghorashi et al. (2014) have investigated the 2nd and 3rd generation and describes the problems (Muslim) immigrants face and the strategies that are employed to make headway on the labour market. They find that higher-educated immigrants often emphasize ‘sameness’ with their professional colleagues from native parentage in order to build confidence and trust. At the same time, immigrants prefer to hold on to parts of their identity and tradition which can grant them a ‘sense of belonging’ and act as a reservoir for emotional support. This, they argue, results in a ‘balancing act’ in which immigrants are constantly caught up trying to bridge different group
25
boundaries (Alba, 2005). Works such as these give us a more direct impression of the challenges and difficulties faced by (Muslim) immigrants and their descendants.
Protestants stand out as having the highest likelihood of being employed, something which is sometimes ascribed to their ‘work ethos’. However, this might also be explained by the general wealth of Northern Protestant Europe vis-à-vis the Catholic South. We have tried to account for wealth discrepancies by including our GDP indicator in this model, but the gaps still remain suggesting that religion has a separate effect.
Contextual effects?
For our GDP variable we used a continuous variable ranging from around 13000 (Poland) to 93000 (Norway). We realized that in order to properly understand the effects of GDP on employment we had to use a categorical variable as a continuous variable diminishes and obfuscates the observed effects. When using a categorical variable17 we were able to magnify the effects of GDP on employment and found that respondents (including immigrants) living in high GDP countries were more likely to be employed. In other words, wealthy countries offer better employment opportunities to immigrants as well as natives. This seems intuitively not that surprising as wealthier countries are often associated with more business and education opportunities for its inhabitants.
We are unable to verify the hypothesized ‘scarring effect’ posited by Pedace et al. (2012). In our results we found a significant positive relation between GDP growth and the employment status of natives. However, our immigrant population shows mixed and insignificant results, therefore we are unable to draw general conclusions about the influence of GDP growth on employment status. It might be that GDP growth is initially most beneficial for natives who are positioned to reap the most rewards from an economic upswing whereas immigrants are more precariously situated and are unable to take
advantage of new possibilities.
We related our indicator ‘Social Spending as % of GDP’ to the typology originally presented by Esping-Anderson (1990). We hypothesized that countries with large amounts of public social spending18 (i.e. Social-Democratic states) would spend more on integration efforts and in turn are more likely to increase the participation of immigrants. It does seem from our results that immigrants originating in refugee sending countries fare better in countries with larger social expenditures, which could be interpreted as a sign that some of these expenditures are beneficial for the economic integration of our refugee population.19 Taken as a whole, it does not seem that immigrants or natives living in countries with larger social expenditures fare better or show higher labour market uptake. The relationship between public expenditures and the socioeconomic participation of immigrants is an interesting but politicized topic which we think deserves more extensive and elaborate research.
17
Results on our categorical GDP variable can be found in the appendix (No. 4)
18
Social spending is defined by OECD as expenditures on social protection, but also on active labour market policies, for more info see:
http://stats.oecd.org/OECDStatDownloadFiles/OECDSOCX2007InterpretativeGuide_En.pdf
19
26
Our last contextual indicator was concerned with the attitude of the domestic population vis-à-vis immigrants. We hypothesized that positive attitudes vis-à-vis immigrants will translate into more opportunities and easier access to the labour market. Negative attitudes, we hypothesized, would obstruct and reduce integration into the labour market. Our results do not seem to point in this direction. Attitude towards immigrants does not seem to be a strongly associated with employment status. Nonetheless, attitudes might in the long-term impact educational attainment or other social indicators such as psychosocial development, which in turn does impact labour market status. There might be an indirect relationship between these determinants. As we are particularly concerned with 1st generation immigrants who often arrive during adolescence or later it might be that attitude is of less importance, but the importance of attitude might me more pronounced for 2nd and 3rd generation respondents.
To conclude for our contextual determinants, we only found GDP to show a clear unidirectional relationship with the labour-market status of immigrants. Our other contextual variables do not show strong associations, larger public expenditures are not associated with increased immigrant labour market participation, except for immigrants originating in what we termed ‘refugee’ sending countries.
Measuring the effects of Policy
We have arrived at our most challenging section, measuring the influence of policy (i.e.
legal-institutional context) on the labour market participation of immigrants. MIPEX country scores have been imported and recoded into our main ESS database. Some of these policies are only relevant for a specific immigrant subpopulation. For example, our policy variable ‘access to the Public Sector’ is only relevant for respondents who do not hold EU citizenship. Therefore, we had to make sure to exclude internal EU migrants in our analysis.20 The same holds true for our policy variable “Immediate Access to the labour market”. This is only relevant for our immigrant population who are recent arrivals.
Our policy variables display strong collinearity, countries which score high on one of the previously summarized indicators have a strong tendency to score high on other policy indicators as well. Therefore we decided to create separate policy domains and to make three separate models. We restricted our models to the immigrant population and included individual and economic contextual indicators to make sure that our results are not just indicators of wealth but are actually giving us a sense of the effects of policy and the legal-institutional framework within which immigrants operate.
Shown below is a table of our results:
20
27
Table 5. Logit analysis on main dependent variable for separate policy domains; including individual and contextual determinants.
28
Policy: Access Policy: Study Policy: Security Variable
Education: Primary (Ref.) (Ref.) (Ref.)
Lower secondary 0.22* 0.21* 0.23* Upper secondary 0.53*** 0.54*** 0.55*** Vocational 0.71*** 0.78*** 0.73*** Tertiary 0.94*** 0.96*** 0.96*** Gender: Female -0.62*** -0.62*** -0.62*** Language 0.23*** 0.23*** 0.21*** Age 0.29*** 0.29*** 0.29*** Age^2 -0.00*** -0.00*** -0.00*** Cohabiting/Marriage 0.26*** 0.25*** 0.26*** Essround: 4 Round 5 -0.18* -0.14* -0.15* Round 6 -0.27*** -0.23*** -0.25*** Contextual-level: GDP 0.00*** 0.00 0.00*** GDP Growth 0.26*** 0.19*** 0.14*** Policy-level:
Immediate LB access after arrival
Neither (Ref.)
Family or Labour Migrants -0.22
Both -0.02
Access to Public Sector Employment
None (Ref.)
Partial 0.25
Yes 0.62***
Access to Study Grants
None (Ref.)
Partial -0.26**
Yes 0.07
Facilitated Recogniton of Diploma's & Skills
Weak (Ref.)
Partial 0.49***
Strong 0.41***
Access to Social Security
Weak (Ref.) Partial -0.34** Full 0.08 Constant -5.80*** -5.81*** -5.51*** N 7679 7679 7679 Pseudo r^2 0.11 0.12 0.11
