Economic crisis and occupational integration of recent immigrants in Western Europe

Abstract

Spanish

French

There are two models of inclusion of recent immigrants in Western Europe. In the Continental model immigrants are penalized in terms of both probabilities of being employed and job quality. In the Mediterranean one there is a stronger trade-off between a limited risk of unemployment and a lower job quality. Did the recession foster a convergence or a divergence between these two models? This article focuses on this issue and analyses the integration of immigrants in 10 countries, using EU-LFS data (2005–2012) and considering two occupational outcomes: the probability of being employed, and the probability of avoiding the unskilled working class. It also studies the turnover between unemployment and dependent employment. The crisis generated a partial and limited convergence between the two models, involving only male immigrants living in Southern Europe. In these countries immigrants experienced higher risks of unemployment because the crisis diminished their turnover between unemployment and dependent employment.

Keywords

Economic crisis Europe immigration/migration

Introduction

The global economic crisis had important effects on international migration, for instance on incoming flows, on return migration, on sending of remittances to the country of origin and, of course, on the employability of immigrants (Tilly, 2011). This article focuses on the occupational integration of recent immigrants in Western Europe, analysing how the features of the Southern European and the Central-Northern models of inclusion changed with the onset of the economic crisis (Ballarino and Panichella, 2015a; Reyneri and Fullin, 2011). Did the recession foster a process of convergence among European countries, or did it exacerbate the peculiarities of the two models of inclusion?

This article seeks to answer this question by comparing, during the period 2005–2012, the occupational status of recent immigrants from Eastern Europe, Asia, Africa and Latin America and natives in 10 Western European countries: Greece, Italy, Portugal, Spain, Belgium, France, Germany, Netherlands, Sweden and the UK. This comparative and diachronic design, including both old and new receiving countries belonging to both models of inclusion, makes it possible to study the labour market integration of immigrants over time in order to observe how it changed with the crisis. Moreover, the empirical analyses systematically differentiate male and female populations, highlighting how the inclusion of immigrants changes according to gender (Ballarino and Panichella, 2017), and how the economic downturn had different effects on male and female immigrants.

Empirically, this work comprises two analyses. The first compares immigrants and natives by looking at the changes over time of two occupational outcomes: the probability of being employed and of avoiding the unskilled working class. The aim of this part is therefore to investigate how the crisis affected the ‘ethnic (or migration) penalty’, i.e. the disadvantages of immigrants when socio-demographic characteristics are controlled for (Heath and Cheung, 2007), and how its effects differed among Western European countries. In other words, the ‘ethnic penalty’ measures whether immigrants received lower occupational returns with respect to natives with similar individual characteristics, such as age, education, family status and region of residence. Empirical researches often considered this concept as a ‘black box’, since most of available data do not allow us to disentangle specific mechanisms of disadvantages: scant transferability of human capital, discrimination practices, social network, and so on. The second analysis tries to uncover how the penalty concerning the probability of being employed changed during the crisis, examining two different risks: (a) the probability of losing the current job; (b) the probability of (re)finding a new job during the unemployment spell. Hence, the second part of the analysis studies the difference between immigrants and natives in regard to these transitions, shedding light on the possible mechanisms underlying the effects of the crisis on the ethnic penalty.

The article is organized into six sections. After this introduction, the second section describes the characteristics of the inclusion of recent immigrants in the European labour markets, while the third section outlines specific research hypotheses on the effects of the economic recession. The fourth section presents the data, variables and techniques used in the analysis, while the fifth sets out the empirical results. Finally, the sixth section presents the study’s conclusions.

Immigrants in the European labour markets: The effects of the crisis

Different models of integration

The empirical literature has analysed the ‘ethnic penalty’ in the labour market by considering two occupational outcomes: the risk of unemployment and the quality of the job (Ballarino and Panichella, 2015a; Reyneri and Fullin, 2011). On the basis of the combination of the penalty in these two outcomes, it is possible to systematize the models of inclusion of immigrants in the labour markets into four ideal-types according to the typology reported in Table 1.

Table 1.

Ideal-types of immigrants’ integration in the labour market.

		Job quality
		High penalty	Low penalty
Employment	Low penalty	1. Trade-off	3. Integration
Employment	High penalty	2. Double penalty	4. Selection

The first ideal-type (quadrant 1) is characterized by a trade-off between unemployment and job quality, where immigrants have similar (or even higher) probabilities of being employed than natives, but at the same time are strongly penalized concerning the quality of the job. The second quadrant comprises the double penalty type, where immigrants are penalized in both occupational outcomes. The third quadrant is the integration one, which is characterized by a low penalty in both occupational outcomes. This type includes, for instance, the integration of skilled immigrants from other Western countries, whose employment condition is in general similar to that of the native population (Ballarino and Panichella, 2015a). Finally, the last type is the selection one, where immigrants are penalized regarding the risk of unemployment, but once they become employed (i.e. they pass the selection process), they face a limited penalization concerning job quality. This outcome may depend on both the labour market structure, which limits the entry of immigrants but does not penalize them in quality jobs, and the selectivity of immigrants themselves in terms of aspirations, skills, ambitions, etc.

As stressed in the introduction, the focus of this work is the labour market integration of recent immigrants from Asia, Africa, Eastern Europe and Latin America. Following the previous literature (Ballarino and Panichella, 2015a; Reyneri and Fullin, 2011), this study excludes the ‘old’ immigrants, who often migrated within the recruitment programmes organized by most of the Centre-North European countries until the 1970s (Reyneri and Fullin, 2011), whose integration into the labour market took place in a different historical and economic context. The labour market integration of the ‘old’ immigrants, indeed, can be included in the first type, since their risks of unemployment were relatively low, because of the guest-worker programmes, but they were mainly confined in the unskilled blue-collar job positions of Fordist manufacturing (Panichella, 2014).

By contrast, the occupational paths of recent immigrants among Western European countries are more heterogeneous, and their integration can be traced to two different models: the Mediterranean (or Southern European) model and the Central-Northern (or Continental, or Northern) one (Reyneri and Fullin, 2011). Both models are defined by the labour market structure and regulation of the host countries, which determines the different features of the ethnic penalty (Ballarino and Panichella, 2015a). In sum, empirical studies have shown that a strict employment protection legislation increases insider–outsider cleavage in the labour market (Lindbeck and Snower, 1988), reducing the employability of immigrants (Kogan, 2007), while the flexibility of the labour market indirectly promotes entrance into the labour market, although at the lowest levels of the occupational hierarchy (Ballarino and Panichella, 2015a).

The Mediterranean model can be included in the first quadrant of the typology reported in Table 1, because it is characterized by a strong trade-off between the risk of unemployment and the job quality. In Italy and in Spain, as well as in Portugal and Greece, immigrants and natives have similar unemployment rates, but the segmentation of the Southern European labour markets pushes foreign workers into the lowest strata of the occupational structure, reducing their opportunities of career mobility (Ballarino and Panichella, 2015a; Bernardi et al., 2011; Fullin and Reyneri, 2011). This also applies to immigrant women who, despite their relatively high employment rate, especially when first-migrants (Ballarino and Panichella, 2015b, 2017), are often trapped in unskilled and unstable employment positions in the domestic and personal care sector, where they play a valuable role within familistic welfare regimes (Anderson, 2000). The Southern European trade-off between unemployment and job quality closely depends on the prevalence of small firms and on the large underground economy (Reyneri, 1998, 2001), where employment is regulated in an extremely flexible way and its extensive demand for low-skilled labour can hardly be met by the native workforce (Reyneri and Fullin, 2011), especially in a context of limited internal geographical mobility (Panichella, 2014).

Also in the Central-Northern model, the allocation of recent immigrants occurs in the secondary labour market, which includes more volatile economic sectors with unskilled and unstable occupations (Heath and Cheung, 2007; Kogan, 2004; Piore, 1979). For instance, in countries such as Germany or Denmark there is the skill barrier (Esping-Andersen et al., 1994), whereby the regulations concerning qualified manual jobs require a vocational training qualification, which is unlikely to be possessed by immigrants, even if they have the substantial skills required. However, differently from the Mediterranean countries, in Central and Northern Europe immigrants are strongly penalized also in terms of risk of unemployment.

Once employed, immigrants living in Central and Northern Europe are also less likely to have a good job than the native population, although this ethnic penalty is lower than the one faced by immigrants living in Southern Europe, and it tends to diminish over time of residence in the host country (Ballarino and Panichella, 2015a). This is, therefore, the second characteristic that differentiates the Central-Northern model from the Mediterranean one. Hence, the labour market integration of recent immigrants in the countries of Central and Northern Europe has the characteristics of the double penalty ideal-type, where foreign workers have more difficulty in finding jobs than natives and are also penalized in regard to job quality.

The Central-Northern model includes both Central European and Scandinavian countries, where the features of the inclusion of (new) immigrants are quite similar. Also the UK can be included, with some strain, in this model. However, in the UK the labour market is regulated in a flexible way, and there are weak industrial relations. This makes the inclusion of immigrants peculiar, since the unemployment of immigrants is likely to be less than in other Continental and Northern European countries (Ballarino and Panichella, 2015a; Kogan, 2007; Reyneri and Fullin, 2011).

The Mediterranean and the Central-Northern models have been analysed in static terms, without considering their change over time. The analysis of the economic crisis is therefore interesting because it implies a diachronic perspective, which is a novelty in the existing literature and makes it possible to study whether the two models diverged or converged with the onset of the economic downturn, or if differences within the two models emerged (Fellini, 2017).

The following paragraphs point out the possible effects of the crisis on both models, while the third section describes the analytical strategy and the empirical expectations.

The effects of the crisis

It is difficult to establish a priori whether the economic downturn fostered a divergence or a convergence between the two models: the economic crisis may have had different effects, even conflicting ones, on the employment condition of immigrants (Ambrosini and Panichella, 2016). As regards the first dimension of the ethnic penalty, namely the risk of unemployment, there are reasons to expect that the effects of the crisis have been stronger among immigrants than among natives. Immigrants are often included in the volatile ‘secondary’ labour market, where employment is highly variable over the course of the business cycle, while natives are mostly employed in the ‘primary’ labour market, which is less sensitive to economic recessions (Kogan, 2007). Moreover, most male immigrants are included in some specific employment sectors, such as construction and labour-intensive manufacturing, which are more affected by economic cycles and have been strongly hit by the crisis (Awad, 2009). This negative effect of the crisis may be even stronger in the Mediterranean model, where immigrants are often concentrated in small and micro enterprises, where it is easier to be fired, and in the irregular black labour market, where a normally high turnover may have been exacerbated during the recession (Ambrosini and Panichella, 2016).

Other arguments, however, go in the opposite direction. First, while most male immigrants are employed in sectors which are deeply affected by the economic conjuncture, foreign women are overrepresented in the domestic and personal care sector, where the labour demand is less sensitive to the trend of the economic cycles and appeared to suffer less in the recession period (Awad, 2009). A similar argument applies for those male and women immigrants working in the agricultural sector, where the labour demand has been only partially affected by the crisis.

Besides the patterns of the sectors in which immigrants are employed, also some of their individual behaviour may have limited the negative effects of the downturn. Immigrants have a sort of ‘instrumental approach’ to job search, especially in the early stages of their settlement in the host country. They aim at minimizing unemployment spells and at maximizing the short-term economic returns to their geographical mobility, often looking forward to the possibility to return to their country of origin. This is why immigrants prefer to enter the labour market quickly, even at the cost of entering the lower occupational strata, where demand for labour is constant, rather than to invest time and resources in a long search for a higher-status job (Dustmann, 2000). Because of the flexibility of the irregular labour market, such a pragmatic job search strategy is particularly widespread in Southern Europe, thus increasing immigrants’ employment probability, albeit in the lowest strata of the occupational structure.

While the crisis may have had mixed effects on the risk of unemployment, with regard to job quality – the second labour market outcome that we consider – it is reasonable to expect a persistent disadvantage of immigrants, or even an increasing deterioration of their penalty. From this point of view, the instrumental approach to job search is likely to have a negative effect, since the search for short-term returns may induce immigrants to accept any job, thus increasing their risk of being trapped in the most unstable and unskilled jobs. Second, even when immigrants are able to reduce their risk of unemployment, as when they obtain jobs in agriculture or personal care, their chances of leaving the lowest strata of the occupational structure are generally limited, and it is possible that the economic downturn has further reduced such opportunities. It is difficult to determine whether these dynamics were strongest in the Mediterranean or Continental/Northern countries. In both models, the job quality of immigrants is systematically lower than the average quality of natives’ jobs, although there are important differences in the magnitude of this gap. However, since a low job quality for most immigrants is a structural feature in European labour markets, we expect that with the onset of the economic downturn this occupational outcome remains stable or increases further.

Besides the analysis of the two occupational outcomes that differentiate the two models of inclusion, this study also looks at the effect of the crisis on the transitions from unemployment to dependent employment, and vice versa. The analysis of these outcomes, which are often neglected by the literature (Reyneri and Fullin, 2011), makes it possible to study the ‘qualitative’ differences of the ethnic penalty concerning the probability of being employed, and how they changed with the crisis.¹ In some countries, for instance, more immigrants can be unemployed than natives because they are more often fired, while in other countries their penalty can be the consequence of a greater risk of remaining in unemployment longer, because it is more difficult to find a new job during unemployment spells.

The magnitude of this turnover is different between the Continental and the Mediterranean models. Since the transitions between dependent employment and unemployment is especially widespread in the black and irregular labour market, where firing costs are lower and job conditions are set through informal agreements (Ambrosini and Panichella, 2016), immigrants living in Southern Europe have higher probabilities of being fired than the native population, but at the same time they have greater opportunities to find a new (unskilled and/or unstable) job when unemployed. This turnover is at the basis of the lower ethnic penalty concerning employment opportunities that characterizes the Mediterranean model of inclusion (Fullin, 2011; Silva and Vázquez-Grenno, 2010). In Southern Europe, indeed, immigrants rapidly enter the labour market, but they are mostly confined in the lowest strata of the occupational structure, where they enter the most unstable and unskilled job positions and experience higher risks of being fired. In the Continental countries, by contrast, turnover costs are higher, even at the lowest levels of the occupational structure; for this reason, immigrants experience higher penalties when moving from unemployment to dependent employment (Kogan, 2004). Moreover, strict employment protection should increase the insider–outsider cleavage and segregate many job losers into unemployment because employers adopt stricter hiring practices (Reyneri and Fullin, 2011). For these reasons, we expect that the economic downturn reduces the turnover of immigrants between unemployment and dependent employment more in Southern Europe than in Northern and Continental Europe, thus increasing the penalty in the Mediterranean model of inclusion.

We now move to the analytical strategy of the article, and to a set of research hypotheses derived from the above discussion.

Comparison strategy and empirical expectations

As stressed above, the empirical analyses of this article are divided into two parts: first, it studies how the crisis affected the ethnic penalties concerning the probability of being employed and the quality of employment; second, it looks more specifically at the probability of moving from dependent employment and vice versa.

The analyses focus on recent migrants, defined as those who have emigrated for less than 10 years, living in 10 European countries belonging to both the Mediterranean and the Central-Northern models of inclusion. Greece (GR), Italy (IT), Portugal (PT) and Spain (ES) are included in the Mediterranean model, while Belgium (BE), France (FR), Germany (DE), the Netherlands (NL), Sweden (SE) and the United Kingdom (UK) are part of the Continental one.

As argued above, there are strong differences in how the labour market is regulated, which determine the patterns of the ethnic penalty concerning the risk of unemployment and the quality of the job. France, Germany, Belgium and Sweden are effectively coordinated market economies (Hall and Soskice, 2001), where strict labour market regulation imposes high firing costs on employers. This can lead the latter to stronger statistical discrimination and thus to a systematic preference for hiring native instead of immigrant workers (Kogan, 2007). Italy, Spain, Greece and Portugal are also coordinated market economies, but they have a huge and flexible underground economy which favours migrants’ access to employment (Ballarino and Panichella, 2015a). Finally, the UK is in between these two groups of countries: although it is part of the Central-Northern model, the labour market flexibility of its liberal market economy (Hall and Soskice, 2001) makes inclusion in the labour market easier (Kogan, 2007). Also the Netherlands has specific characteristics: between the 1980s and the 1990s several measures were taken to allow for more flexible labour contracts in the Netherlands (Teulings and Hartog, 1998), making the integration of immigrants living in that country more similar to that of those living in the UK than in Germany or France (Ballarino and Panichella, 2015a).

Regarding the effect of the crisis on the two models, three possible scenarios can be theorized. The stability scenario predicts that the impact of the crisis on the difference in occupational integration between natives and immigrants is similar in all countries considered, leaving unchanged the differences between the Southern European and the Central-Northern models. By contrast, if the effect of the crisis on the ethnic penalty differs between the two models, there can be either a convergence or a divergence scenario. The convergence scenario predicts that the differences between the two models of inclusion became smaller with the onset of the economic crisis, making the patterns of immigrants’ occupational integration more similar across European labour markets. This may be due, for instance, to a greater increase in the risk of unemployment in Southern Europe, where before the crisis immigrants had employment rates similar to (if not higher than) those of natives. The alternative scenario is the diverging one, where the peculiar characteristics of the two models of inclusion strengthened with the economic crisis. This may have happened because in Southern Europe immigrants further increased their probability of being employed (although in low-skilled jobs), or because in Central-Northern Europe the penalization concerning employment increased more than in Southern Europe. Similarly, there may also be a diverging scenario if the penalization of job quality (conditioned to employment) has grown more in the Mediterranean countries.

Data, variables and methods

Data and variables

We used the European Union Labour Force Survey (EU-LFS) data set (2005–2012), limiting the sample to individuals aged 25–60. After missing cases were excluded, the analytical sample size included 9,371,594 individuals (8,613,831 natives and 757,763 immigrants, defined as below). Descriptive statistics of EU-LFS data are reported in the Appendix.

The main independent variable was immigrant status, distinguishing native population from those born abroad. Since the integration of the ‘old’ immigrants in Western Europe is completely different from that of the ‘new’ ones, the two models of inclusion were analysed considering the most recent immigrants (Ballarino and Panichella, 2015a; Reyneri and Fullin, 2011), defined as those living in the host country for less than 10 years. Older immigrants were included in a ‘residual’ category, whose results are not shown. Migrants were divided into four categories according to their geographical area of birth: (a) North America and Australia; (b) EU15; (c) Eastern Europe; (d) Africa, Asia and Latin America.² Since the first two categories have peculiar characteristics and their employment condition is in general similar to that of the native population, in the description of the empirical evidence we will focus only on those migrants coming from Eastern Europe, Africa, Latin America and Asia. Lack of data did not allow to consider more detailed definition of geographical origin, hence the concept of ‘ethnic penalty’ has been broadened to all immigrants from these regions, without considering further important distinction among ethnic groups, as in the case of Andean and other Latin American immigrants with Spanish or Portuguese origin (Reher and Requena, 2009). However, we have performed a set of robustness checks in order to control whether the heterogeneity among different groups might bias our results (see later).

The ethnic penalty was measured by means of two variables. The first was employment status, coded 1 for the employed (dependent or self-employed) and 0 for both unemployed and inactive. The second was job quality, which took the value 1 if the respondent was employed outside the unskilled working class, as defined by the Erikson–Goldthorpe–Portocarero (EGP) class scheme (IIIb, V-VI-VIIab, Erikson and Goldthorpe, 1992), and 0 if s/he was employed in the unskilled working class. Educational level was the first control variable, coded into three categories: up to lower-secondary education (ISCED 0–2); upper-secondary education or post-secondary non-tertiary education (ISCED 3–4); and tertiary education (ISCED 5–6). Models also controlled for age (five-year groups from 25–29 up to 55–60) and marital status (divorced/widowed, single or married).

Methods

As mentioned above, the empirical analysis was divided into two parts. The first analysed the effect of the crisis among native and immigrants concerning the probability of being employed and the probability of avoiding the unskilled working class. It did so by means of linear probability models (henceforth, LPM) with robust standard errors.³ The beta coefficient of LPM can indeed be directly interpreted as marginal effect, limiting the problems of comparability of the beta coefficient among different logit models (Mood, 2010).

The basic model was as follows:

\begin{array}{l} E (Y | X) = α + β_{1} (IMM \times YR) + β_{2} EDU + β_{3} MARSTAT + β_{4} (AGE) + \\ β_{5} (REG) + ε \end{array}

(1)

The model estimated the association between geographical origin $(IMM)$ and the two occupational outcomes, controlling for own education $(EDU)$ , marital status $(MARSTAT)$ , age $(AGE)$ and region of residence $(REG)$ , at NUTS2 (Classification of Territorial Units for Statistics) level. The impact of the crisis was studied by means of an interaction effect between immigrant status and year of the survey $(YR)$ .

Between-country variation was observed by estimating the model separately for each country, but the analyses were also replicated using a pooled data set with an interaction between migration status and country of residence, with similar results (available on request from the author). Models for job quality were estimated both on the whole sample (unconditional model) and only on the employed (conditional model), but for lack of space only the conditional estimations are reported.

The second part of the analyses focused on the turnover between unemployment and dependent employment, which might explain the different characteristics of the ethnic penalty among countries. The analysis of this turnover was based on the retrospective question concerning the labour market condition one year before the survey contained in the EU-LFS. However, the EU-LFS does not report the retrospective information concerning job quality, for this reason this work does not analyse the transition out of the unskilled working class. Since in the EU-LFS the information concerning the geographical origin is not available before 2005, in this analysis the time axis started from 2006 instead of 2005.

We analysed this turnover by looking at two occupational transitions: (a) the probability of being employed as a dependent at t₁ among individuals unemployed at t₀; (b) the probability of being in unemployment at t₁ among dependent employees at t₀. Each transition was estimated with an LPM with the same specification as described in equation 1.

Finally, since the two models of inclusion were analysed in a ‘relative way’, comparing the occupational positions of immigrants with that of the native population, the figures presented in the next section report the beta coefficient estimated from LPM. These coefficients measure the differences in the predicted probability that Y = 1 between immigrants and natives (reference category).

A set of robustness checks were performed. First, other measures of job quality were used, including different measures of low-quality jobs (including all non-permanent jobs; occupations with an International Standard Classification of Occupations 1988 (ISCO-88) score equal to or higher than 8; occupations with ISCO-88 equal to 9). Second, analyses were replicated with a set of Heckman sample selection models, estimating simultaneously the probability of being employed and the probability of avoiding the unskilled working class. Third, analyses were replicated by separating different immigrant groups (Africa, Asia, Latin America and Eastern Europe).

Results are shown in graphical form. Figure presented in the next section do not show the 95% confidence intervals, whose inclusion would make representation of the results less clear. However, in the text only statistically significant differences are commented on. The complete outputs of the statistical analyses are available on request.

Empirical evidences

Figure 1 reports the migrants’ penalty in the probability of being employed and in the probability of avoiding the unskilled working class among the male population.

Figure 1.

Male population: Gap in the probability of being employed and avoiding the unskilled working class by year of survey and country (linear probability models, beta coefficients).

The results clearly confirm the peculiarities of both models of inclusion. The line depicted with a black circle (employment penalty) is systematically higher in the Mediterranean countries, apart from the dramatic reduction of Spain and in Greece in recent years, confirming that in these countries immigrants face a relatively lower penalty in the probability of being employed compared with those living in Continental and Northern Europe. Also confirmed is that in all the countries considered immigrants are strongly penalized in terms of job quality: the lines depicted with a diamond shape, which measure the gap in the probability of avoiding the unskilled working class, are always lower than 0, although with different magnitudes. The UK in the pre-crisis period is an exception to this general pattern, since the inclusion of recent immigrants had the features of the ‘selection’ ideal-type (Table 1): on one hand, they are penalized regarding the probability of being employed but, once they ‘pass’ the selection and enter employment, their penalization concerning job quality is lower or even null.

Empirical results also show that the effects of the economic crisis varied among countries. In Northern/Continental Europe it had a limited effect on ethnic penalty, with the exception of Sweden, where the disadvantage of immigrants concerning job quality significantly increased up to −0.22 in 2012. A similar trend was found in the UK, where the penalty concerning the probability of avoiding the unskilled working class rose from 0.04 in 2005 to −0.11 in 2012, making the inclusion of immigrants more similar to the ‘double penalty’ ideal-type. In Southern Europe, by contrast, immigrants and natives had similar probabilities of being employed before the crisis, but after 2008 the risk of unemployment significantly increased in all Mediterranean countries. Spain is a prime example in this regard, since the gap between immigrants and natives was −0.1 (p = .004) in 2005 and then grew from −0.06 to −0.15 between 2008 and 2009, with a dramatic increase of 9 percentage points. In Greece immigrants were even at an advantage in the probability of being employed before the crisis (as in Italy), but after 2008 their penalization increased up to −0.10 in 2012. At the same time, in Greece we observe a reduction of the penalty concerning job quality, which is probably due to the greater selection of immigrants who managed to find a job during the crisis. Also in Italy, before the crisis immigrants had a higher probability of being employed than native-born Italians, while at the end of the observation window, in 2012, they were slightly disadvantaged (−0.02, p = .000). Finally, in Portugal the employment penalty increased (from −0.02 in 2005 to −0.04 in 2012, p = .000), mainly because a number of immigrants were employed in the agricultural sector (Peixoto et al., 2009), which was less affected by the economic recession.

The general features of the two models of inclusion are clear also when the female population is analysed (Figure 2): also in this case the lines (those depicted by black circles) of the Mediterranean countries are higher than those of Continental countries and Northern Europe, while the lines concerning job quality are always lower than 0. Moreover, it also interesting to note that the migration penalties are generally stronger among women than among men (Figure 1), and particularly high are those concerning job quality in Southern Europe.

Figure 2.

Female population: Gap in the probability of being employed and avoiding the unskilled working class by year of survey and country (linear probability models, beta coefficients).

As expected, the crisis had different effects among women. In this case the employment penalty remained roughly constant in most countries, with the exception of Spain, Greece and Portugal, where it significantly increased after 2008. However, in these three Mediterranean countries the increase of this penalty is lower than that observed for male immigrants, confirming that the recession had a minor impact on the demand for personal care services, in which the majority of immigrant women in Southern Europe are employed.

The lower increase of the ethnic penalty regarding employment occurred at the cost of an increased segregation of immigrant women in the lower strata of the occupational structure. Indeed, in most countries immigrant women faced a higher penalization concerning job quality after 2008. The strongest increase was in Italy, where the penalization of immigrant women was −0.32 in 2005 and rose to −0.38 in 2008 and −0.48 in 2012.

To highlight how the models of inclusion changed with the crisis, Figures 3 and 4 show, for male and female populations respectively, the average magnitude of both ethnic penalties in the pre-crisis years (2005–2007) and the post-crisis ones (2009–2012). For both genders two LPMs were estimated: the first considered the pre-crisis period, controlling for survey years (dummies from 2005 to 2007); the second considered the post-crisis period, controlling for year of survey dummies (2009–2012). Since the crisis started in 2008, but with different speeds across countries, we compare the pre- with the post-crisis period excluding 2008.

Figure 3.

Male population: Gap in the probability of being employed and avoiding the unskilled working class in the pre- and post-economic crisis period (linear probability models, beta coefficients).

Figure 4.

Female population: Gap in the probability of being employed and avoiding the unskilled working class in the pre- and post-economic crisis period (linear probability models, beta coefficients).

Figures 3 and 4 show four ideal-types of models of inclusion of immigrants described in Table 1, which are outlined according to two cut-points. The black lines (penalty equal to 0) differentiate the cases in which immigrants are at an advantage (β > 0) or an disadvantage (β < 0) on the two occupational outcomes. The dashed lines (β > or < 0.05) aim to make the boundaries among the different ideal-types of inclusion of immigrants more ‘blurred’, since immigrants from Eastern Europe, Africa, Asia and Latin America are in general penalized in the labour market, although with different magnitudes among European countries. Also in the Mediterranean model, indeed, they have a similar (or even slightly higher) probability of being employed than the native population, but there is no clear advantage of immigrants with respect to natives. For this reason, if we only use the black lines as cut-points, all countries will be included in the same quadrant, thus limiting the description of the heterogeneity of different models of inclusion.

Figures 3 and 4 show that the crisis fostered a partial and limited convergence process. It was partial because it involved only the male population living in Southern Europe and limited because it only concerned the probability of employment. Indeed, concerning this occupational outcome, the labour market integration of male immigrants in the Mediterranean countries becomes more similar to the one that characterizes the Northern and the Continental countries. Looking at Figure 3, it is in fact clear that in the pre-crisis period the difference between the countries was more mixed, while in the post-crisis period all countries were concentrated in the double penalty type.

It is also interesting to note that the process of convergence has affected not only the Mediterranean countries but also the UK, which before the crisis was in the selection type, while after 2008 it was also included in the double penalty quadrant.

Among women (Figure 4), by contrast, the stability scenario has much empirical support, since in this case the Southern and the Continental clusters are clearly visible and differentiated even in the post-crisis period: on one hand, the four Mediterranean countries are included in the trade-off type, while on the other hand all the Northern and Continental countries are clustered in the double penalty type.

At this point it is necessary to study the mechanisms underlying this partial and limited convergence process, to check whether it is associated with the changing probability of any specific occupational transition. As mentioned above, the differences between immigrants and natives in the probability of being employed can be the result of two outcomes: (a) a different probability of losing the current job; or (b) a different probability of finding a new job during the unemployment.⁴ Figures 5 and 6 focus on the differences between immigrants and natives in the probability of having these occupational transitions, distinguishing the male and female populations.

Figure 5.

Male population: Occupational transitions (linear probability models, beta coefficients).

Figure 6.

Female population: Occupational transitions (linear probability models, beta coefficients).

Figure 5 focuses on the male population and shows that the turnover between dependent employment and unemployment changes markedly among European countries. It is relatively limited in Northern and Continental countries, where immigrants have a higher probability of moving from dependent employment to unemployment, and a lower probability of being hired when they are unemployed (Kogan, 2004). Also in Southern Europe male immigrants have a higher risk of unemployment than natives, but at the same time they also have more chance of finding a new job when they are unemployed.

Regarding the effects of the economic downturn, in Northern and Continental countries the differences in the turnover rate between immigrants and natives remain constant, with the exception of the Netherlands, where between 2007 and 2008 there was an increase, albeit contained (+3 percentage points, p = .000), in the disadvantage regarding the risk of being fired from the dependent labour market. On the other hand, in all Mediterranean countries the ethnic penalty concerning the transition from dependent employment to unemployment increased after 2008, while the probability of moving from unemployment to dependent employment became closer to that of natives. This pattern is very clear, again, in Spain, where the advantage of male immigrants in the possibility of moving from unemployment to dependent employment strongly diminished from 0.15 in 2007 to 0.05 2008 to close to zero in 2012 (0.01, CIs: −0.02; 0.03). On the other hand, the ethnic penalty concerning the risk of being fired in the Spanish dependent labour market increased from 0.03 to 0.08 between 2008 and 2009. It is also interesting to note that the crisis had a stronger effect on immigrants as regards the possibility of finding a job when unemployed, while the disadvantage regarding the risk of losing a dependent job position rose less. This means that in Southern European countries labour markets have contracted at their lower levels, where most of immigrants are included, reducing their chances of re-entry in the dependent labour market during an unemployment spell.

The different turnover between unemployment and dependent employment in the two models is confirmed also among women (Figure 6). However, differently from the male population, here the trends remained unchanged with the progress of the crisis. Partial exceptions are Spain and Portugal, where there is a reduction of the turnover after 2008, although it is smaller than that experienced by male immigrants living in those two countries. In general, the findings of these analyses are in line with the stability scenario for women: since with the rise of the economic downturn there is no clear increase in the penalization concerning transitions from dependent employment to unemployment and vice versa.

To sum up, these analyses also confirm a partial and limited convergence process among countries, because after the crisis we observed a reduction among immigrants of the turnover between dependent employment and unemployment in Southern Europe, where before the crisis they were much more mobile than the native population. This reduction, moreover, mainly involved the possibility of finding a new job during unemployment, since the crisis reduced the employment possibilities in those sectors and occupations that are more unstable and thus more affected by economic downturns.

Conclusion and discussion

This work has studied the effects of the recent economic crisis on the inclusion of immigrants in the labour market, comparing 10 European countries with different labour market regulation. The main aim has been to study how the features of the Mediterranean and the Central-Northern models, which characterize the inclusion of recent immigrants in the European labour markets, changed with the rise of the economic recession.

The empirical results did not show a generalized deterioration of the occupational conditions of immigrants. On the contrary, the negative effects of the economic downturn were limited and partial, since they involved specific occupational outcomes and varied according to gender and to the host country of residence. The effects of the crisis were stronger among males, since most of them were employed in sectors strongly affected by the economic downturn, like construction and labour-intensive manufacturing. As a consequence, with the progress of the economic crisis they experienced a stronger penalization concerning the probability of being employed, without any improvement in job quality. This pattern is particularly clear in Southern Europe, where immigrants concentrate in the irregular labour market, unstable job positions and in small businesses, where exposure to the risk of unemployment is higher.

The results are different for immigrant women. Although also in this case there is a clear difference between the Continental and the Mediterranean models of occupational integration, the effect of the crisis was similar in all the countries considered. The economic downturn had a limited effect on their probability of being employed, since most of them were able to rely on a stable – or even increasing – labour demand in the domestic and care work sector, but at the same time it increased their risk of being segregated in the lower occupational strata.

The second part of the empirical analysis focused on the turnover between unemployment and dependent employment. The aim was to detect in more depth the mechanism at the basis of the (partial and limited) convergence process which involved the Mediterranean model of inclusion. Because of the high flexibility of the lowest strata of the labour market, before 2008 in Southern Europe both male and female immigrants were at greater risk of unemployment than natives, but at the same time they also had more chance of finding a new job when unemployed. After the crisis, by contrast, there was a reduction of both transitions, thus diminishing the employability of immigrants. It is also interesting to note that the crisis had a stronger impact on the probability of (re)finding a new job during unemployment than on the risk of losing a dependent job. This suggests that the increased ethnic penalty in the Mediterranean countries is not due to a greater risk of exiting from the dependent labour market (because of a dismissal or conclusion of a fixed-term job), but rather to lower probability of re-entry in the dependent labour market during employment. This pattern was particularly pronounced among immigrant men, while for women the impact of the crisis on turnover was limited. In Central and Northern Europe such turnover remains almost unchanged, since immigrants suffer a double disadvantage: they have more probability of losing a dependent job and, at the same time, less probability of finding a new job when unemployed.

We think that future research on this topic will progress in four directions. First, the increased segregation of immigrant women must be analysed more thoroughly. Results suggest that in a context of limited job opportunities, immigrant women tend to remain employed, even at the risk of increasing their segregation in the lower levels of the occupational structure, while native women may exit (even temporarily) from the labour market, devoting more time to family duties because their need to obtain supplementary income is lower. This difference can have important consequences, for instance on fertility behaviour and, in the long run, also on the educational integration of the second generation. Second, it is necessary to consider entry into self-employment, which may be an effective strategy for immigrants to deal with the employment difficulties or limited career opportunities that they faced during the crisis. In this regard, the decision to enter self-employment may have mitigated the negative effects of the economic crisis. Moreover, it is also possible that the transition to self-employment differs among immigrants with different educational levels. Third, results concerning the risk of remaining in unemployment should be replicated using panel data, since this occupational outcome cannot be properly analysed in a cross-sectional setting. Indeed, the results reported in Figures 5 and 6 show the difference in the probability of remaining unemployed for two consecutive years, while it is possible that interesting differences between immigrants and native arise when longer unemployment spells are taken into account.

Finally, the effect of the crisis on return migration should be analysed in depth. On this point, several arguments predict a limited effect of the crisis. Since the recent crisis had a global impact (Alexander, 2010), also involving the countries of origin of major migratory flows, return migration may have become riskier than staying in the host country, even if employment opportunities in destination countries have fallen substantially. Indeed, the relationship between economic cycles and return home is not so obvious because return migration is a territorial displacement, and as such involves risks and costs, especially for undocumented immigrants (Rendall et al., 2010). It is thus necessary to analyse the issue of returns with more informative data, in order to control for possible biases related to this selection process.

Footnotes

Appendix

Table A2.

Native population and immigrants from Eastern Europe, Asia, Africa and Latin America by educational level (percentages).

Native population
	Lower	Upper	Tertiary	Total	(N)
GR	38.7	38.7	22.6	100.0	986,700
IT	45.3	40.7	14.1	100.0	2,233,233
PT	73.7	13.6	12.7	100.0	578,017
ES	49.5	19.8	30.7	100.0	622,159
BE	26.8	38.0	35.2	100.0	352,117
FR	26.6	44.4	29.0	100.0	1,302,430
DE	11.3	61.0	27.7	100.0	547,441
NL	24.9	42.9	32.2	100.0	548,773
SE	35.9	35.4	28.7	100.0	836,539
UK	27.4	40.3	32.4	100.0	358,830
Immigrants from Eastern Europe, Asia, Africa and Latin America
	Lower	Upper	Tertiary	Total	(N)
GR	51.1	36.4	12.4	100.0	88,444
IT	47.8	41.3	10.9	100.0	180,105
PT	50.8	28.1	21.1	100.0	36,534
ES	41.8	34.0	24.2	100.0	42,064
BE	43.0	29.6	27.4	100.0	42,692
FR	44.5	29.1	26.4	100.0	139,164
DE	46.2	37.2	16.6	100.0	36,624
NL	36.5	39.6	24.0	100.0	46,842
SE	38.2	23.1	38.7	100.0	102,255
UK	24.9	41.3	33.8	100.0	43,039

Acknowledgements

Author thanks Gabriele Ballarino for their valuable comments on a previous version of this work.

Funding

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

Notes

Author biography

Nazareno Panichella is Assistant Professor of Economic Sociology at the University of Milan. His research interests include social stratification and mobility, migration, family and labour market dynamics.

References

Alexander

(2010) The impact of the economic crisis on the world’s poorest countries. Global Policy I(1): 118–120.

Ambrosini

Panichella

(2016) Immigrazione, occupazione e crisi economica in Italia. Quaderni di Sociologia 72: 115–134.

Anderson

(2000) Doing the Dirty Work: The Global Politics of Domestic Labour. London: Zed Books.

Arango

Baldwin-Edwards

(eds) (2014) Immigrants and the Informal Economy in Southern Europe. Abingdon: Routledge.

Awad

(2009) The Global Economic Crisis and Migrant Workers: Impact and Response. Research Paper No. 433612. Geneva: ILO.

Ballarino

Panichella

(2015a) The occupational integration of male migrants in Western European countries: Assimilation or persistent disadvantage? International Migration 53(2): 338–352.

Ballarino

Panichella

(2015b) Condizione occupazionale e dinamiche familiari delle donne immigrate in Italia. Quaderni di Sociologia 67: 83–106.

Ballarino

Panichella

(2017) The occupational integration of migrant women in Western European labour markets. Acta Sociologica. Epub ahead of print 29 August 2017. DOI: 10.1177/0001699317723441

Bernardi

Garrido

Miyar

(2011) The recent fast upsurge of immigrants in Spain and their employment patterns and occupational attainment. International Migration 49(1): 148–187.

10.

Dustmann

(2000) Temporary Migration and Economic Assimilation. IZA Discussion Paper, 186. Bonn: IZA.

11.

Erikson

Goldthorpe

(1992) The Constant Flux: A Study of Class Mobility in Industrial Societies. New York: Oxford University Press.

12.

Esping-Andersen

Rohwer

Sørensen

(1994) Institutions and occupational class mobility: Scaling the skill barrier in the Danish labour market. European Sociological Review 10(2): 119–134.

13.

Fellini

(2017) Immigrants’ labour market outcomes in Italy and Spain: Has the Southern European model disrupted during the crisis? Migration Studies. Epub ahead of print 27 September 2017. DOI: 10.1093/migration/mnx029

14.

Fullin

(2011) Unemployment trap or high job turnover? Ethnic penalties and labour market transitions in Italy. International Journal of Comparative Sociology 52(4): 284–305.

15.

Fullin

Reyneri

(2011) Low unemployment and bad jobs for new immigrants in Italy. International Migration XLIX(1): 118–147.

16.

Hall

Soskice

(eds) (2001) Varieties of Capitalism: The Institutional Foundations of Comparative Advantage. Oxford: Oxford University Press.

17.

Heath

Cheung

(2007) Unequal Chances: Ethnic Minorities in Western Labour Markets. Oxford: Oxford University Press.

18.

Hellevik

(2009) Linear versus logistic regression when the dependent variable is a dichotomy. Quality and Quantity 43(1): 59–74.

19.

Kogan

(2004) Last hired, first fired? The unemployment dynamics of male immigrants in Germany. European Sociological Review 20(5): 445–461.

20.

Kogan

(2007) Working Through Barriers. Dordrecht: Springer.

21.

Lindbeck

Snower

(1988) The Insider-Outsider Theory of Employment and Unemployment. Cambridge, MA: MIT Press.

22.

Mood

(2010) Logistic regression: Why we cannot do what we think we can do, and what we can do about it. European Sociological Review 26(1): 67–82.

23.

Panichella

(2014) Meridionali al Nord. Migrazioni interne e società italiana dal dopoguerra a oggi. Bologna: Il Mulino.

24.

Peixoto

Sabino

Abreu

(2009) Immigration policies in Portugal: Limits and compromise in the quest for regulation. European Journal of Migration and Law 11(2): 179–197.

25.

Piore

(1979) Birds of Passage. Cambridge: Cambridge University Press.

26.

Reher

Requena

(2009) The National Immigrant Survey of Spain: A new data source for migration studies in Europe. Demographic Research 29(12): 253–278.

27.

Rendall

Brownell

Kups

(2010) Declining return migration from the United States to Mexico in the late 2000s recession. RAND Population and Labor Working Paper, WR-720.

28.

Reyneri

(1998) The role of the underground economy in irregular immigration to Italy: Cause or effect? Journal of Ethnic and Migration Studies 24(2): 313–331.

29.

Reyneri

(2001) Migrants’ Involvement in Irregular Employment in the Mediterranean Countries of the European Union. International Migration Papers, 39. Geneva: ILO.

30.

Reyneri

Fullin

(2011) Labour market penalties of new immigrants in new and old receiving West European countries. International Migration 49(1): 31–57.

31.

Silva

Vázquez-Grenno

(2011) The ins and outs of unemployment and the assimilation of recent immigrants in Spain. Journal of Population Economics 24(4): 1309–1330.

32.

Teulings

Hartog

(1998) Corporatism or Competition? Labour Contracts, Institutions and Wage Structures in International Comparison. Cambridge: Cambridge University Press.

33.

Tilly

(2011) The impact of the economic crisis on international migration: A review. Work, Employment and Society 24(4): 675–692.