I. INTRODUCTION

Migrants as a percent of the world’s population have observed a steady rise over the last fifty years, growing from 2.3% in 1970 to 3.3% in 2015. According to the World Migration Report (2018), approximately 3% of the global population comprised international migrants in 2015. Nations worldwide have witnessed the growing mobility of individuals for various economic, social and security reasons, making migration a prominent policy issue for governments, institutions and people around the world. Matters such as integration, displacement, safe migration and border management have emerged as significant concerns for all States today, considering the ever-expanding movement of individuals across and within land. With more than 740 million people having migrated within their own country of birth (as per the most recent global estimate), it is safe to conclude that it isn’t only international migration that has increased over the years, but internal migration too has undergone significant changes.

The increasing feasibility of international movement can be attributed to various factors such as digital revolution, distance-shrinking technology and reductions in travel costs. As migration dynamics continue to evolve with time, it is imperative that resources be diverted towards understanding and studying international migration and its various manifestations to ensure that the varied and diverse need of migrants is met in their respective host countries, in addition to attention extended towards natives by the host countries’ government and institutions.

One of the most tangible and significant ways in which immigration impacts host countries is through its effect on the domestic labor market of that host country. Immigrants, with their own characteristics (e.g. abilities, language, culture, etc.), present the natives with an opportunity to function with complementary units in the labor market, providing significant value additions. However, entry and exit to the labor market for immigrants brings with it its own set of hardships and challenges.

Since labor market outcomes for immigrants also depend on factors independent of skill, capability and qualification, various studies have been undertaken to evaluate and assess the determinants of participation of immigrants in the labor market of their host countries. In our paper, we have aimed to explore the relationship between an immigrant’s ethnic identity and their closeness or lack thereof to it, and how their respective status stands to affect their role in the labor market.

Ethnic identity is defined as the sense of self-identification with a group, society, culture or a setting of the country of origin. (Constant & Zimmermann, Work and money: payoffs by ethnic identity and gender. In Ethnicity and Labor Market Outcomes , 2009) Having a commitment or considering one as a group is beneficial in aspects like networking, resilience to changes and self-esteem. However, these benefits almost disappear with migration, as ethnic groups become a minority in host countries, where they often face discrimination. (Phinney & Ong, 2007) They are discriminated against even though author like Constant and Zimmermann (2009) concluded that immigrants with their own characteristics, like abilities, language, and culture, function as significant value additions as they complement natives. The OECD publications have corroborated the status of some immigrants by showing there is a gap between employment of natives and immigrants in European countries, principally for those immigrants who belong to Non-EU countries. (OECD , 2014) This leads us to raise the following to question: does having a strong ethnic identity causes immigrants to be less attractive to the labor market?

The objective of this study is to examine the relationship between ethnic identity and unemployment of non-European immigrants in the period 2002-2016 using an instrumental variable approach. Before running our model, we perform a principal component analysis to construct an ethnic identity index and then use linear and non-linear regressions to evaluate different relations among our variables of interest. In the following section we summarize the literature review surrounding this topic. Section 3 describes the data and empirical methodology used. Section 4 presents the main results, and section 5 presents the conclusion.

II. LITERATURE REVIEW

According to Akerlof and Kranton (2000), identity is a person’s sense of self bound to social categories, wherein individuals identify with people in some categories and differentiate themselves from those in others. Identity is therefore associated with social environments and expected respective behavior, deviations from which can generate disutility. Social categories include racial and ethnic designations, following which we can describe ethnic identity as an extent to which members of an ethnic group associate themselves with their ethnic background or culture. Since an immigrant’s ethnic identity can affect his/her decision to work, we intend to analyze how, and the corresponding impact that creates on the dynamics of the host country’s economy.

One of the main issues with ethnic identity is its measurement. Caselli and Coleman (2013) proposed an economic model to formalize the notion of endogenous ethnic identity as the concept was gaining popularity among social scientists in the early 2000s. Their model included resources being allocated based on the ethnic composition of a society, and individuals being free to choose their identity strategically. They could switch ethnicity by paying a cost, which would increase with greater physical or cultural distance among groups. Their attempt to formalize this concept bore important implications for building relevant measures of ethnic diversity in the future. Recently, Constant and Zimmermann (2008) in their article titled Measuring Ethnic Identity and its Impact on Economic Behavior advocated for a new measure of ethnic identity of migrants, modelling its determinants and exploring its explanatory power for economic performance. They constructed a measure labelled ethnosizer to identify the intensity of a person's ethnic identity, which was built from information on language, media, ethnic self-identification, ethnic networks, and residency plans. The ethnosizer was a two-dimensional concept, classifying migrants into four states namely, assimilation, integration, marginalization, and separation. The ethnosizer was found to mainly depend on pre-migration characteristics, and exogenous to economic activity. Their work concluded that ethnic identity significantly affected economic outcomes. Pendakur et al. (2005) Defined ethnic identity as attachment to the group who share one’s ancestral heritage and used a direct measure of the depth of ethnic identity based on a survey question which asked, “is your ethnic origin very important to you, somewhat important, not very important or not at all important”. This measure was used to explore how ethnic identity affected two labour market processes often presumed to depend on it-- the use of informal networks to find jobs, and the quality of jobs found. The paper concluded that for European ethnic minorities the depth of ethnic identity was positively correlated with the propensity to use informal methods to find jobs, a result consistent with ethnic identity playing a role in ethnically based job-finding networks. It was also observed that for minority men, ethnic identity was visibly correlated with much lower occupational prestige, a result consistent with ethnic identity being correlated with behaviour that serves to separate ethnic minorities from majority populations. In this study, we base our measure on the work done by Alberto Bisin et al (2010).

Another issue on the analysis of ethnic identity is that its relationship with labor market is in general endogenous. Islam and Raschky (2015) recognized that consistent estimates of the effect of immigrants’ identity on labour market outcomes were complicated by the endogenous relationship between performance on the labour market and attitudes towards ethnic identity. Their paper instead used a measure of genetic distance between an immigrant’s home and host countries as instrument for an immigrant’s identity. They found evidence for adverse effects of home country identity on male immigrants’ unemployment likelihood and that a stronger host country identity only has a systematic effect on employment and job satisfaction. Their work however concluded that overall, immigrants’ identity appears to play only a negligible role in their labour market performance.

There are studies of ethnic identity in the European Union. Bisin, Patacchini, Verdier and Zenou (2011) studied the relationship between ethnic identity and labour market outcomes for non-EU immigrants in Europe. Their work concluded that there was a penalty to be paid for immigrants with a strong identity, first generation immigrants paying a penalty of about 17% while second generation immigrants having a probability of being employed not statistically different from that of natives. However, second generation immigrants with strong identity had a lower chance of finding a job than natives. Their analysis also revealed that type of integration and labor market policies in host countries were important factors on which the relationship between ethnic identity and employment depended. Their result also indicated that more flexible labour markets helped immigrants to access the labour market but didn’t protect those who had a strong ethnic identity.

Also, there are different approaches to analyse the effects of ethnic identity on the labour market. One way is to look at discrimination in recruitment by doing experiments with fictitious applicants. Blommaert, Tubergen and Coenders (2012) studied the effects of explicit and implicit interethnic attitudes on ethnic discrimination in hiring. Their paper explained that unlike explicit attitudes, implicit attitudes were characterised by reduced controllability, awareness or intention. They used data from a laboratory experiment consisting of an Implicit Association Test, a questionnaire and a recruitment test in which participants reviewed résumés representing fictitious applicants who varied basis ethnicity, gender, education and work experience. The results they obtained showed that only explicit interethnic attitudes affected discrimination in grades.

Other way of looking at the effects of ethnic identity is by analysing real data on the employer-employee relationship. Åslund, Hensvik, and Skans (2014) investigated how the interplay between manager and worker origin affected hiring patterns, job separations, and wages. They built numerous specifications using a longitudinal matched employer-employee database including 70,000 establishments, which consistently showed that managers were substantially more likely to hire workers of their own origin. Their results concluded that workers who shared an origin with their managers earned higher wages and had lower separation rates than dissimilar workers, and that this pattern was driven by differences in unobserved worker characteristics. Their findings indicated that sorting patterns were more likely to be explained by profit-maximizing concerns than by preference-based discrimination. Gorinas (2014) investigated the relationship between employment and ethnic identity, complementing existing literature by capturing a fresh dimension of ethnic identity--openness to majority norms. Reproducing measures from earlier studies, she found that immigrants’ employment outcomes did not systematically associate with their ethnic identity but that immigrants who shared social norms with the majority experienced significantly better employment outcomes, particularly first-generation immigrant women. Her study also showed that interethnic differentials in majority norms could account for up to 20% of the explained part of the employment gap between natives and first-generation immigrants.

Kesler and Safi (2017) used Labour Force Surveys to empirically examine inequalities that immigrants of disadvantaged minority origins faced in labour force participation, employment, and earnings in France and the U.K. The UK labour market was found to attract immigrants who had a larger skill advantage over natives though barriers to labour force participation were high. In France, however, barriers to employment among active job-seekers were more important. Earnings inequalities were however found to be less significant in both countries and barriers to opportunity largely similar for immigrants of disadvantaged minority origins. We observe inequalities of strikingly similar magnitude in the two societies, despite overarching differences in labour market organization and immigrant integration policies.

It is therefore evident that an individual’s ethnic identity stands to play an influential role in their decision to work in their respective host county. Since the intensity of ethnic identity can differ between individuals from different groups, in between men and women, and also within a given group for different individuals, it is an important field of study for researchers and academicians.

In this study we define the intensity of an individual’s ethnic identity for which we develop an index basis language spoken at home, religious commitment and traditions followed and use that to study its impact on the individual’s participation in the labour market of the country they immigrated to. This study not only holds important implications for job-seekers but also for job-providers. Discrimination, both implicit and explicit, and distinguished social and cultural effects of how integrated an individual is in their host country’s economy in relation to their ethnic identity is what drives our interest in assessing the impact of this on the various host countries’ economies.

III. DATA AND METHODOLOGY

Data

We use database from the European Social Survey (ESS). This survey is carried out every 2 years and has been conducted since 2002 with updated information available up to 2016. Therefore, we have data for eight rounds used in our study. The population under study is the first and second generation of Non-EU migrants in EU countries. The variables we consider for our study are based mainly on Bissin et al. (2011)

The ESS has a special section of immigration questions, which includes information on origin, relatives and attitudes towards immigration. Our main variable to predict is unemployment represented in the database as a dichotomous variable that takes value 1 if the individual is unemployed and 0 if he/she is not. The predictor variable "ethnic identity" has been constructed as an index composed of 3 indicators: "language" which is an indicator if the main language spoken at home is different from the official language of the country where he/she resides, "tradition" which is a categorical variable that takes values ​​from 1 to 6, 1 being the score individuals assign if they do not consider important to follow tradition or customs from where they come from and 6 that they consider it very important. It is relevant to mention that this variable has the order changed since the values ​​were inverted. Finally, the "religious" indicator is a categorical variable that assigns values ​​from 0 to 10; 0 implying that the person is not religious and 10 as very religious. Additionally, we applied weights to the calculations, which was recommended in the methodological guide of the survey.

The variables used, and their corresponding meaning are as follows:

Immigrant: Non-EU nationals for whom one parent’s country of origin is not in the EU

First generation: Immigrant who is not born in EU (especially the hosting country). It is a binary variable coded 1 if the immigrant is a first generation and 0 elsewhere.

Second generation: Immigrant born in the host country and with one parent from outside the EU. It is a binary variable coded 1 if the immigrant is a second generation and 0 elsewhere.

Unemployment: We consider those who are unemployed and actively looking for a job. It is a binary variable coded 1 for unemployed and 0 if not.

Ethnic identity index: It measures how an individual is attached to his/her ethnic group. We construct it using principal component analysis (PCA) conducted with three variables: attachment to religion, attachment to tradition and language most spoken at home.

Attachment to religion: It is an ordinal variable coded from 1 (not at all religious) to 6 (very religious)

Attachment to tradition and customs: It is an ordinal variable coded from 1 (not at all) to 6 (very much)

Language most spoken at home: A binary variable coded 1 if they do not talk the language of the host country at home and 0 if they do.

Other relevant variables

These are variables we will use in our empirical strategy as presented below:

- Years since arrival in the host country

- Education level from 1 to 5, where 1 is less than lower secondary education and 5 is tertiary education

- Gender

- Age

Construction of the index

Before constructing the index, we tested and generated regressions between the three components and our variable of interest "unemployment"; these regressions were made in several ways. We tested them separately first-each variable with the dependent variable, and then joint regressions with all possible combinations and new variables grouped as non-linear combinations. Controls were added to ensure the significance of the variables that comprised the index. Additionally, transformation of the original scales were made for each variable to ensure the best specification of the model that estimates the significance of each of them.

Once the tests were done, we selected our three variables described in the previous section and decided to use the variables in their original scale, except for "tradition", whose values ​​were inverted, so the lower values ​​represented less attachment to customs and tradition. It is important to mention that the variable "tradition" turned out to be insignificant to explain "unemployment", however we decided to maintain it to build the index because it generated a more consistent index that better explained the dependent variable and lays in sync with the existing literature on the importance of maintaining the traditions of the country of origin on the labor market in the host country.

After this, the index was constructed using the principal component analysis method "PCA". Under this criterion, the component that collected the greatest variability of the data was selected (approximately 50%). The scoring coefficients show that the index is almost equally driven by the variables "religious" and "tradition"; both variables collecting between themselves approximately 80 percent of the variability of the data, while "language" the remaining 20 percent. Because the calculated index had negative values, to make the interpretation of the results easier, we converted the index values ​​to a scale between zero and one.

Methodology

In this section, we test our main hypothesis derived from our literature review on ethnic identity. We check whether ethnic identity is causing the feature observed in labor outcomes.

For this purpose, we considered the following regression specification:

The left-hand-side variable $y_{ijt}$ is whether Non-EU immigrant $i$ in country $j$ at time $t$ is unemployed or not. ${EthId}_{ijt}$ is the ethnic identity index, a continuous measure of how an immigrant is attached to his/her ethnic group. ${X_{ijt}}_{}$ represents a set of covariates that we consider in our model to control any other observables that may explain the labor market outcome, such as education, age, and years since arrival.

${\mathrm{\theta }}_t$ represent time fixed effects to capture any time variability (different rounds of the survey), ${\mathrm{\gamma }}_j$ are host country fixed effect to count for any domestic relevant information like facts about immigrant and the labor market regulation, and ${\eta }_{co}$ are the country of origin fixed effects. ${\epsilon }_{ijt}$ is the error term.

We attempt first to estimate this model as a linear probability model, to measure the extent to which labor market outcomes are explained by ethnic identity.

Empirical challenges and solutions

We face a causality issue if we stick to the OLS estimate. Doing so, we face some empirical challenges regarding the exogeneity of the ethnic identity index. When an immigrant is facing a repulsive immigration policy or experiences difficulties to enter the labor market, this can affect his ethnic identity. So being unemployed can affect the ethnic identity as well. This a reverse causality problem.

Our solution is to find an instrument for ethnic identity index. We consider the average of the value of the index for immigrants whose country of origin is the same and are living in different host countries as an instrument for the ethnic identity index for a given immigrant living in a specific host country of EU. As the average calculated is country of origin specific characteristic for the immigrant, it was to provide an exogenous variation of the index and not affect the labor market outcomes in other ways but only through the index.

Ethnic identity is a set of elements that is not homogeneous in all countries, that is, it cannot implicitly assume the ethnic identity of an immigrant from a Muslim country is on average equal to one of a Latin American country since the difference in culture positions regarding how the labor market works in both geographic spaces would not reflect the true effect of the index on the labor market (i.e.,, the role of women in the labor market of a Muslim country is not the same as that of a Latin American country and even within Latin America there are notorious differences between countries).

Another consideration is whether we consider a linear probability model (OLS) or we consider a non-linear one (a probit model) as our dependent variables are dichotomous. So, we use both. We do a linear estimation (OLS and IV) and a non-linear estimation (probit).

IV. RESULTS

In the table below we show that the estimate by ordinary least squares generates a significant result at a confidence level of 99%, that is, the positive correlation between unemployment and the index of ethnic identity is confirmed. We use the instrument to test for causal relationship. We prove that in the first stage of the IV regression the instrument has fulfilled the condition of relevance. However, in the second stage we find that the estimated value of our variable of interest is not significant, so in the context of our specification of the model we cannot say that there is a causal relationship.

As we mentioned in the methodology section, we also used non-linear probability models to test our relation of interest. We can observe from the table that all the non-linear models present a positive relation between ethnic identity and unemployment. In both cases, the coefficient is significant.

Our conclusion about the causal relation between the variables of interest vary depending on the specification considered on the linearity of the probability model. That can be seen from the fact that the IV probit presents a significant coefficient and the IV does not.

As a further analysis, we tested for differences between gender in the construction of ethnic identity. As it is shown below, the variables, used for the index in the case of females, are not significant. While, for men are significant at 90% and 99% level of confidence. These results could be indicative of the fact that gender-based discrimination trumps race/ethnicity-based discrimination.

V. CONCLUSIONS

Ethnic identity is an important determinant in an individual’s participation status in the labor market of the host country they have immigrated to. With growing cross-country mobility, it is therefore imperative that we study the movement of labor across nations to better evaluate the changing dynamics of labor markets as immigrants are introduced into the system. Our results lead us to a positive relation between ethnic identity and unemployment, highlighting that individuals with a stronger sense of self rooted in their home country face roadblocks in their chances of procuring employment in host countries. An individual therefore might strategically choose how well he/she integrates or isolates him/herself from his/her respective host country’s social and cultural identity to better or worsen their opportunity of gaining employment in the latter. Although closeness to ethnic identity for women yields insignificant results, for men their closeness to their ethnic identity plays a larger role.

Our results yield similar conclusions to those obtained in Bisin and Zenou (2010) and Pendakur and Pendakur (2005). However, there is need to continue research in this field and develop stronger models with better indicative specifications to generate results closer to how these factors interact with each other in real time and space. Governments, institutions and people around the world need for this subject area to be better developed to ensure that rights and opportunities of natives and immigrants are both well maintained and protected.

REFERENCES

Akerlof, G., & Kranton, R. (2000). Economics and identity. The Quarterly Journal of Economics, 115(3), 715-753.

Åslund, O., Hensvik, L., & Skans, O. (2014). Seeking similarity: How immigrants and natives manage in the labor market. Journal of Labor Economics, 32(3), 405-441.

Battu, H., & Zenou, Y. (2010). Oppositional identities and employment for ethnic minorities: Evidence from England. The Economic Journal, 120, F52–F71.

Bisin, A., Patacchini, E., Verdier, T., & Zenou, Y. (2011). Ethnic identity and labour market outcomes of immigrants in Europe. Economic Policy, 26(65), 57-92.

Blommaert, L., Van Tubergen, F., & Coenders, M. (2012). Implicit and explicit interethnic attitudes and ethnic discrimination in hiring. Social science research, 41(1), 61-73.

Caselli, F., & Coleman II, W. (2013). On the theory of ethnic conflict. National Bureau of Economic Research, 11, 161–192.

Constant, A., & Zimmermann, K. (2008). Measuring ethnic identity and its impact on economic behavior. Journal of the European Economic Association, 6(2-3), 424-433.

Constant, A., & Zimmermann, K. (2009). Work and money: payoffs by ethnic identity and gender. In Ethnicity and Labor Market Outcomes . Emerald Group Publishing Limited, 3-30.

Gorinas, C. (2014). Ethnic Identity, Majority Norms, and the Native-Immigrant Employment Gap. Journal of Population Economics,, 27(1), 225-250.

Islam, A., & Raschky, P. (2015). Genetic distance, immigrants’ identity, and labor market outcomes. Journal of Population Economics, 28(3), 845-868.

Kesler, C., & Safi, M. (2017). Immigrants in the labour markets of France and the United Kingdom: Integration models, institutional variations, and ethnic inequalities. Migration Studies, 6(2), 225-250.

OECD . (2014). Labour Market Integration of Immigrants and their Children: Developing, Activating and Using Skills. Paris: International Migration Outlook 2014 - OECD Publishing.

Pendakur, K., & Pendakur, R. (2005). Ethnic identity and the labour market. Vancouver Centre of Excellence.

Phinney, J., & Ong, A. (2007). Conceptualization and measurement of ethnic identity: Current status and future directions. Journal of counseling Psychology, 54(3), 271.

Notes