Are refugees better together? a study on the influence of ethnic clusters and dispersal policy on refugee integration

doi:10.21203/rs.3.rs-3090819/v1

Download PDF

Research Article

Are refugees better together? a study on the influence of ethnic clusters and dispersal policy on refugee integration

https://doi.org/10.21203/rs.3.rs-3090819/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

This paper investigated whether ethnic clusters play a role in refugee integration across different races. There are two hypotheses for this study, first, refugees in districts with higher ethnic clusters have better refugee integration, and second, therefore, the dispersal policy is expected to harm refugee integration. I analyzed individual-level data from the Socio-Economic Panel (SOEP), a wide-ranging longitudinal study of private households in Germany, and district-level demographic and economic data from the German statistics department. In this paper, both perspectives of random or non-random refugee allocation are considered while investing the influence of ethnic clusters on refugee integration. A quasi-experimental approach, difference in differences (DID) analysis, is used as the benchmark model to study the causal effect of the refugee influx. In addition, as a robustness check, propensity score matching in the difference-in-differences setup is used to overcome the non-random factor of refugee allocation issued by certain scholars. The results showed that Arabs are better at economically and socially integrating into areas with higher refugee concentration, whereas Asians are worse off. However, ethnic clusters showed a negative relationship but had no causal impact on integration for African and European refugees.

Development Economics

Behavioral Economics

refugees

integration

policies

local environment

economic

social

political

race

religion

ethnic concentration

In general, refugees should not remain refugees forever; either their situation at home improves and they return, or they find a lasting solution in their country of refuge. The latter requires refugees to integrate into the host community. Integration is a vital and intricate two-way process requiring contribution from both parties; refugees should adapt to the host society without foregoing their own cultural identity, whereas host communities and institutions to welcome refugees and meet the needs of the diverse community. The sudden influx of refugees in 2015-16, the European refugee crisis resulted from several interrelated wars in the Middle East; the Libyan civil war, the Syrian civil war, and the war in Iraq resulted in most European countries having a sizeable refugee population likely to remain amid failing to integrate economically and socially. Along with employment, barriers to citizenship and residential and social segregation remain in these countries. The leaders in Europe have struggled to integrate the growing number of refugees seeking asylum across European borders. In Germany, approximately a million refugees crossed the borders seeking asylum from war and persecution in their home country. Refugee integration is a complex and multi-dimensional construct, as integrating refugees into health, educational, economic, social, and political contexts guides how smooth refugee integration occurs (Gordon,1964). The current literature revealed the inadequacy of the integration process in Germany, from immigrants receiving lower employment opportunities (Kogan, 2011) and wages (Constant and Massey, 2003) compared to locals. However, social networks among refugees and immigrants were found to alleviate their economic disadvantages (Dustmann et al., 2016, Kalter and Kogan, 2014). For instance, ethnic clusters or ethnic concentrations are used as a proxy for immigrants’ social networks (Battisti et al., 2022; Damm, 2009; Edin et al., 2003; Kristiansen et al., 2021). The endogeneity between social networks and labor market outcomes is high (McPherson, Smith-Lovin and Cook, 2001; Mouw, 2003). The assumption that social contacts are established non-randomly, the effect of social network effect tends to develop by selection effects. Thus, bias in selection in race or ethnicity creates a major division in social networks (McPherson et al., 2001). The current literature found immigrant inflow to areas related to connections to previously immigrated family or friends (Palloni et al., 2001) or existing co-ethnic groups (Damm, 2009; Tanis, 2020). The studies found that if asylum seekers or immigrants select their location of residence, their decision will likely be driven by family, friends, or co-ethnic networks in different locations. Therefore, social connections might depend on the selection effect by an endogenous choice of networks.

However, European leaders and policymakers have raised concerns about the different ethnic backgrounds of refugees making it further challenging to integrate refugees into the local community and causing segregation among communities. Ethnic segregation occurs for several reasons, such as housing affordability, support networks, and language barriers resulting in a high concentration of a particular group of individuals (i.e., Syrian refugees) living in a particular neighborhood or area. In Germany, reports of ethnic segregation are alarming, as segregation leads to social and economic disadvantages for refugees and local communities, causing cultural enclaves, stereotypes, and political polarization. Thus, making it more important for governments and communities to address the issues of ethnic segregation and create inclusive and integrated societies. After the refugee influx, European policymakers have adapted policies aligning with European migration law to improve integration and avoid ethnic segregation. They have implemented policies and programs to address these challenges and promote integration. These can include language training programs, employment, and education initiatives, and social support services. In 2016, German policymakers passed the Integration act 2016, restricting refugees from migrating to another state for few years, depending on the state-specific adaption of the policy. In addition, states can further restrict refugees from residing in certain municipalities to avoid overcrowding of resources or segregation among communities. The integration act obliges refugees to participate in integration courses which include language and societal knowledge classes, before being granted refugee status (Christ, 2020. Degler and Liebig, 2017; Fontanari, 2022). In addition, the perks and drawbacks of living with their ethnic group can vary on the culture and characteristics of the group itself (Tajfel and Turner, 1979).

On that account, this paper investigates the effect of ethnic clusters on refugee integration by using a difference in difference (DID) analysis strategy. I compared, treatment areas; districts with above-median ethnic refugee share to control areas; districts with below-average median ethnic share, received on average from the year of the refugee influx. This analysis is done for refugees from different ethnic backgrounds. Second, as the policy implied movement restrictions for refugees entering Germany from January 2016 onwards to avoid ethnic segregation, the effect of the Integration Act 2016 on refugees who face movement restrictions, is investigated using interrupted time series (ITS) analysis. The results show that the relationship between ethnic clusters and refugee integration varies across different ethnic refugees. Asians were found to be worse in both economic and social integration measures in areas with greater ethnic share. However, for Asians, the effect was vice versa, as a greater ethnic share resulted in greater employment status and oral language adaption. However, the positive relation turned negative in districts with largest refugee share. Lastly, there was no significant effect for African and European refugees. The ITS results showed that the Integration Act 2016, which restricted refugee movement for a minimum of two years, harmed refugee integration in the short run.

The contribution of this study is threefold. First, the use of the DID analysis allowed control for other factors that might influence integration outcomes, such as differences in economic and social conditions between the districts, isolating the effect of ethnic clusters on integration, and providing a more accurate understanding of the relationship between these factors. Second, as the study investigated the relationship across different ethnic backgrounds, it enriches our understanding of whether the relationship varies among ethnic groups. These findings can provide insight into potential differences due to cultural or social backgrounds and allow policymakers to develop policies accordingly. Lastly, the analysis of the effectiveness of the Integration Act reassures how policies can be implemented to promote integration. The next section of this paper elaborated on the refugee context and scenario in Germany. Section 3 discussed the theoretical underpinning and empirical studies of refugee integration literature. Followed by section 4, which discussed the method and methodology used in this study. In section 5, I discussed the results and analysis and related them to the literature. Lastly, in section 6, the implication and conclusion of this study are discussed.

2.1 Germany and refugees

Germany has a chronicle with refugees, being both an origin and destination for refugees, witnessing different geographic and demographic refugees over the last century. Revisiting world war I, Germany hosted thousands from Eastern Europe, followed by world war II, when millions from war-torn European regions sought asylum in Germany. During the cold war, West Germany hosted refugees from the Soviet Union and Eastern European countries, whereas East Germany was a source of political refugees. In the 1990s, Germany received refugees from south-eastern Europe, the Balkans, due to the Yugoslavian wars. Lastly, the most significant refugee influx was in 2015, when more than 1 million refugees arrived in Germany from war-torn countries such as Iraq, Syria, Afghanistan, and others. According to the Federal Office of Migration and Refugees (BAMF) the refugee influx in the last decade has come from diverse regions, the majority being from the Middle East and Africa, such as Syria, Iraq, Afghanistan, Eritrea, Somalia, and Nigeria. (De Haas, 2008; Türk,1992; Oltmer, 2016). The graph below shows the influx of refugees in 2015, from Arab, Asian, and African regions, whereas a decrease in European refugees.

2.2 Refugee allocation system and the random factor

The first step for asylum seekers is registering with the Federal Office for Migration and Refugees (BAMF). After registration, the refugees are spread across states using a national system called EASY, a quota-based distribution, Königsteiner key. The quota is derived from the tax revenue and population (2/3 of it comes from tax revenues and 1/3 from the population), implying refugees each state receives depend on the state’s economic capacity and population density. However, refugees are allocated to municipalities with no 'one' definite system and are claimed to be random. During the application process, refugees might face movement restrictions depending on the allocated district and individual scenario. For instance, some might require refugees to remain in reception centers where they are provided with basic necessities, whereas some allow private housing. In addition, several incentives for refugees to stay in their designated municipality rather than move to a different district. For instance, first, goods and social services are provided, and (ii) refugees are legally obliged to remain until a decision is made (Gehrsitz and Ungerer, 2022). However, the Integration Act 2016 imposed geographical restrictions for a minimum of 2 years for refugees arriving in 2016, and onwards (UNHCR, 2019). The Federal Office for Migration and Refugees (BAMF) coordinates the allocation of refugees along with the states and municipalities. In current literature, certain scholars have emphasized that housing or job availability might be evaluation criteria for allocating refugees to municipalities. However, the allocation of refugees has been criticized as being very random, causing ongoing pre-issues such as housing shortage, inflation, and unemployment to deteriorate further (Gehrsitz and Ungerer, 2022). The asylum seeker or refugee is not allowed to self-select their location of residence, nor are their preferences taken into consideration during their allocation procedure. In addition, according to the German Migration department, officials responsible for assigning the asylum seeker have no prior background information on the applicant to avoid bias during the allocation procedure. Thus, even if specific criteria influence the district-level allocation of refugees, the quota-based system, with a goal to fairly distribute the burden across districts and states, the allocation can be considered random as refugees have minimum control in deciding their location of residence, and individual characteristics and preferences are not taken into account. For instance, refugees might be allocated to a district, solely based on the quota, but totally different from their preference. Therefore, the allocation of refugees is considered a quasi-random assignment, and the use of the difference in differences (DID) regression analysis for this study is justified.

2.3 Refugee Integration Act 2016

In August 2016, the integration Act and regulations were applied in Germany. This act is a law in Germany that aims to facilitate the integration of refugees into society. The act established the Federal Office of Migration and Refugees (BAMF) to be responsible to oversee refugee integration. This act amended laws regarding integration classes, vocational training, employment and training opportunities, place of residence, and permanent settlement permits for refugees. The goal was to help in the integration of refugees and avoid ethnic segregation across communities in Germany. According to the federal government, this law is to “support and challenge”. For instance, by making it easier and faster for refugees who are eager and determined to integrate. This act obliged asylum seekers who have applied for asylum or entered Germany looking for asylum from January 2016 would be restricted from movement from their allocated locations for a minimum of two years. This is implemented to prevent refugees from moving to specific favorable areas due to employment opportunities or their networks. In addition, states can further restrict refugees from residing in certain municipalities to avoid overcrowding of resources or segregation among communities. The primary reason behind this geographic restriction is to prevent ethnic segregation, which harms integration as refugees.

The field of study on refugee integration is much smaller than the field of migration integration studies. This section discussed the theoretical framework and empirical work on refugee and migratory studies and elaborates on the effectiveness and contribution of this paper.

3.1 Theoretical underpinning

According to social identity theory, an individual’s self-concept is originated from membership in relevant social groups (i.e., Germans, immigrants, local communities), which guide their behavior and relation to other social groups (Tajfel and Turner, 1979). Similarly, social representations theory mentioned how social representations such as ‘refugees’ have values, beliefs, and practices that function as a system of helping people to associate themselves in the social world and make sense (Moscovici, 1973). Both complement each other in social identification, such as being German, immigrant, or refugee is also social representation (Augoustinos, 2001; Wagner et al., 1999).

Integration is when refugees and host communities share resources without conflict (Harrell-Bond, 1986). The integration process can be extended to identity and cultural affiliation during participation (Berry, 1992). Nevertheless, integration as a form of involvement is not unidimensional. Kuhlman (1991) believed refugee integration to be successful when refugees can engage in the host economy while being compatible with local cultural values. The standard of living should satisfy the culturally determined minimum requirements, and the refugees should also be able to maintain their own identity while blending into their new environment. He claimed that a perfectly successful integration would mean the refugee influx would benefit the local community's living standard and economic opportunities. And the difficulties or issues among the two groups should be similar to those previously settled within the host society. Around two decades ago, the United Kingdom (UK) home office commissioned a study on refugee integration indicators. Ager and Strang (2004, 2008) joined to develop a framework for refugee integration. Their framework recognizes ten key domains that withhold understanding of integration. For instance, opportunities and achievements in employment, housing, education, and health guide the rights and terms of refugee integration via three types of social connections; social bridges, bonds, and links. Furthermore, they connected refugee rights and citizenship to work, housing, education, and health outcomes. Lastly, language, cultural knowledge, safety, and stability facilitate integration by reducing the obstacles to inclusion and facilitating two-way exchange between refugees and natives. Researchers have built on this framework and recommended modifications. Hynie, Korn, and Tao (2016) found isolation among refugees from Iran and Afghanistan in the UK while studying relationships between immigration and institutions and communities and suggested trust to be a determinant and thus be included in the framework for integration. Latter, Phillimore (2020) argued the original framework to shift away from individual refugee outcomes to understanding political, relational, and temporal contexts in which integration occurs. Integration can be understood as a multi-dimensional process in which refugees' adaption to each domain is essential for their quality of life. (Borselli & van Meijl, 2020; Phillimore, 2020; Ward et al., 2020). Thus, from a theoretical perspective, I investigated the changes in the refugee and host communities after the influx.

Relating to ethnic concentration and refugee integration, according to social identity theory, if refugees are concentrated in a particular area or community, they might identify themselves and others as a member of the same group. This group identification might influence their relationship and sense of belonging to the new community, thus integrating into the local community. As the theory suggested, individuals’ self-concepts and behavior are influenced by their perception of their own and others’ social status; refugees who perceive themselves as having lower social level than locals might experience difficulty integrating new community, and vice versa.

3.2 Empirical studies

3.2.1 Refugee integration

Refugees finding employment is a significant integration agitator. For instance, jobs are not only important in providing financial and essential support to refugee households; employment is an opportunity to learn the language, build a future, regain confidence, and establish status. The exogenous flow of refugees in the labor market of the receiving country based on the Marshallian factor-demand theory of a competitive labor market found a worsening effect for locals but optimistic on migrant and refugee workers (Borjas and Monras, 2016). A resembling phenomenon was seen in refugee-receiving provinces in Turkey as the refugee influx caused a reduction in Turkish informal labor (Tumen, .2016). The immigrants entering the labor market scale output and reduce costs (Wagner, 2014). Similarly, social integration into local communities is necessary for new migrants or refugees to initiate harmony and peace in society. They can build three primary forms of social capital, bonding, bridging, and linking. These three attributes allow bonding social capital, necessary for social cohesion and support (Woodlcock, 2002; Kawachi, 2006; Poortinga, 2012; Tippens, 2019). Trust and cooperative relationships initiate social bonds between people who share cultural practices and social identities, which are very important in the integration process (Ager & Strange, 2008; Hogg, 2006). Platts-Fowler and Robinson (2015) found that the characteristics of the local population and the mobility of the refugee population can influence the bonding in refugee communities. Lastly, an ideal democracy would mean all constituents have a voice, but it is becoming challenging due to migrants and refugees without citizenship (Hannah Pitkin, 1967). The exclusion of such minority groups inhibits democracy. The inclusion of such groups can not only act as an integration indicator. Still, it should be converted into a goal for societies to increase diversity and equality in the economy, strengthening social cohesion (Ortensi and Riniolo, 2020). Relying on the UN’s Sustainable Development Goals, under the 2030 agenda, the UN member states have agreed to “empower and promote the … political inclusion of all” and to “ensure equal opportunity” as well as to promote “appropriate legislation, policies, and action in this regard” (United Nations, 2021).

3.2.2 Ethnic concentration

The refugees living in an enclave of coethnics or co-nationals harm economic and social integration by making it more difficult for newcomers to acquire the host country’s language and norms (Liebig and Spielvogel, 2021; Schüller, 2022). Braun and Dwenger (2017) found that the similar ethnic share of the population harmed refugees' economic, social, and political integration. For instance, segregated neighborhoods have high poverty levels and lower economic development, harming both the refugees and the host communities. Paul Jargowsky (2002) found that, in the United States, African Americans in neighborhoods with higher concentrations of African Americans had lower employment and higher poverty rates than African Americans who lived in neighborhoods with lower concentrations. His study also found the relationship stronger for African American men than for African American women. A similar phenomenon was observed in the UK; McQuaid and Singh (2015) found that Bangladeshi and Pakistani men living in areas with greater Bangladeshi and Pakistani ethnic groups had lower employment rates than those living in neighborhoods with a lower concentration of people from their own ethnic group. Similarly, when refugees are concentrated in specific neighborhoods or areas, they lack interaction with host community members, leading to social isolation, a lack of social isolation, and an appreciation of cultural differences. Lastly, ethnic segregation might contribute to political polarization within a country, creating integration and social cohesion challenges for governments.

However, the current literature on the relationship between ethnic clusters and integration is bilateral. For instance, Marten, Hainmueller, and Hangartner (2019) found that a higher concentration of coethnics makes it easier to find jobs improving the economic integration of refugees. Work wages improve living standards and reduce the government's burden on social benefits. Concurrently, the efficacy of European dispersal policies was analyzed by Fasani, Frattini, and Minale (2021). They collected data from European Labour Force Survey (EULFS) for countries such as Denmark, Ireland, Netherlands, Norway, and the UK and conducted a difference in difference analysis. Their quasi-experimental study showed that such policies worsened the economic integration of refugees as labor market participation suffered. Furthermore, employment bans and restrictions are vital in harming refugee integration prospects, lasting for ten years after arrival (Fasani, Frattini, and Minale, 2022).

3.2.3 Other factors

There are other factors that influence refugees or migrants to integrate into the local community. For instance, housing availability is a fundamental right for refugees and helps social inclusion (Netto, 2011; Phillips, 2006). Relating to Ager and Strong (2008), housing affects refugees' physical and emotional health. Therefore, governments receiving a large influx of migrants and refugees maintain housing policies and pricing, as the housing shortage raises a barrier to refugee integration, which can cause more significant discrimination and spatial segregation (Landau, 2006). Similarly, education improves integration; schooling provides displaced individuals with socialization and instills hope (Mosselson et al., 2017). A highly educated displaced individual is more likely to be a member and contribute to the new society. Education improves human capital, which positively affects earnings and the standard of living (Lange and Topel, 2006), reassuring the importance of access to education for refugee children integration and avoiding segregation. Therefore, policymakers strive to navigate the school system successfully and provide support to prevent the isolation and exclusion of refugee children in schools. Refugees tend to be exposed to dreadful events for political or religious oppression, war, migration, and resettlement, suffering from physical and mental trauma. For instance, half of the Syrian refugees in Germany were diagnosed with Post Traumatic Stress Disorder (PTSD) (Hajak, Sardana, Verdelli and Grimm, 2021). Therefore, access to health care is mandatory for refugees to integrate into society. In addition, power differential and language barriers between the doctor and the refugee patients also hinder integration (Newbold & McKeary, 2018). Proficiency in the host language is a significant factor in establishing refugees’ sense of belonging to a new society. Language can secure employment and social connections, especially among first-generation migrants or refugees (Kristjansdottir and Christiansen, 2017; Van Tubergen & Kalmijn, 2005). The lack of language proficiency would discourage firms from recognizing the preliminary credential of refugees. Further, the local community would have a negative representation of the migrant or refugee community (Ager and Strang, 2008; Jackson and Bauder, 2014; Tomlinson, 2010).

3.3 Research contribution

According to social identity theory, individuals tend to categorize themselves and others into groups based on shared characteristics. These characteristics, such as nationality, ethnicity, race, or other factors, influence their attitude, perception, and behavior. In the context of ethnic concentration and integration, belonging to specific groups, such as refugees in a concentrated area, would influence behavior and attitude influencing integration. The literature highlighted the complex nature and importance of addressing economic and social factors in supporting refugees to achieve long-term integration in their host societies and investigated the direct relationship between ethnic concentration and integration factors across countries. Relating to social identity theory, the formation and characteristics of the group are related to the group’s perception, attitude, and behavior. Therefore, this study investigated the relationship across Arab, Asian, and European refugees, highlighting whether ethnic differences play a role in this relationship. Lastly, there are limited studies on the effectiveness of the dispersal effect of the Integration Act 2016 on shaping refugee integration.

3.4 Research questions and objectives

Q1. Whether ethnic clusters positively influence the economic and social integration of refugees?

Q2. Whether dispersal policy and refugee integration are negatively related?

The study’s objective is (i) to investigate whether ethnic clusters play a role in refugee integration across different ethnic groups. The 2015 refugee influx has created different areas with diverse ethnic refugee concentrations; thus, comparing these districts provides insight into refugee population integration. (ii) to investigate the effect of the dispersal policy, integration act 2016, on policy-driven movement-restricted refugees.

This section discusses the details of the data sources and the identification strategy and specification used in this paper.

4.1 Data

I have used individual-level data from the Socio-Economic Panel (SOEP) data sets to capture the dependent and independent variables for this study. The dependent variable, economic and social integration, is captured using questions on employment and local language adaption. The data set is a longitudinal survey conducted in the Federal Republic of Germany from 1984 to 2020, produced by Deutsches Institut für Wirtschaftsforschung (DIW), Berlin. The variables include demographic characteristics, employment, health, life satisfaction, attitude, and migration and integration indicators. The SOEP data has refugee samples, which allowed to investigate and compare integration across different ethnic refugee groups. The data used is from the period 2013 to 2020, addressing the refugee influx in 2015, and refugee samples conducted during the mentioned timeframe. The questions used to capture economic integration are whether the individual is employed full or part-time; social integration is whether the individual can speak the local language, German; Second, I collected district-level population and housing prices in square meters from Germany's federal and state statistics departments, and refugee and migration-related data from the German office of Migration and Refugees (BAMF), allowing us to capture the ethnic concentration across all German districts.

4.2 Identification strategy

For the first objective, a simple OLS regression is run to identify the relationship between ethnic clusters and refugee integration while controlling for other covariates that might influence the relationship. Then the causal effect is found, using a difference in differences (DID) analysis. In addition, I address the other perspective regarding the randomness of refugee allocation as robustness check, using propensity score matching in a difference-in-differences setup. For the second objective, as movement restriction is applied to refugees across all districts migrating in 2016, I have used interrupted time series analysis to find the causal relationship between the policy and refugee integration.

4.2.1 Difference in difference strategy

I used the difference in differences (DID) strategy, a quasi-experimental approach to find the causal effect of ethnic concentration on refugee integration. The DID analysis compares changes in integration outcomes pre- and post-the-2015 refugee influx between treatment groups, districts with high, mid, and low ethnic-specific refugee shares, and control groups, districts with corresponding lower ethnic-specific refugee shares. The analysis is run across different ethnic groups Arab, Asian, African, and European refugees (see eqn, 2 and 3). Our model specifications are below,

yidt = α + β1 Treatment area ethnic refugee shared + B2 Postt + B3(Treatment area ethnic refugees shared × Postt ) + Xidt + εidt (1)

In all three models, y_idt represents the binary indicator of refugee participant i’s employment status, and language adaption, living in district d, and interviewed in time t. It was included in the model to control for common trends and shocks to integrate that may correlate with factors that vary permanently by district and systemically change over time. The ethnic refugee share is the concentration of the particular ethnic group compared with the district population. For instance, Arab refugee concentration_d is the ratio of refugees from Syria, Iraq, and Afghanistan and the general population in district d. Asian, African, and European refugee concentration is the ratio of refugees from their respective continents and the general district population. Post_t is a dummy variable for the post-influx period that takes the value of 1 if observations are after the refugee influx in 2015, and 0 otherwise. The coefficient B₃ estimates the average treatment effect. X_idt is a vector of individual covariates that includes the participant’s age, sex, and district-level controls such as housing prices and agricultural employment ratio. The district-level controls are added as literature emphasizes their influence on refugee integration. ɛ_idt is the idiosyncratic error term that captures time-varying unobserved heterogeneity. The treatment areas are identified as districts that received above-average ethnic refugee share on-average from the year of refugee influx. For instance, to find the causal effect of ethnic cluster on refugee integration of Arab refugees, I compared treatment districts, districts that received above average median Arab refugees, to districts below average median Arab refugee share.

In addition, to avoid bias, I have controlled for time-varying control variables in the model, which might capture changes in the characteristics of the treatment and control groups that might affect the outcome. For instance, the refugee share measure considers the population differences at each time period across time. In addition, I have used a fixed-effects model to control the time-invariant difference between our treatment and control groups. Lastly, this study assumes both treatment and control groups have parallel trends, i.e., if the treatment were not implemented, rather than the overall, both groups would have the same trend for employment and language adaption for each group.

4.2.2 Interrupted time series (ITS) analysis

To investigate the influence of the dispersal policy of the Integration Act 2016 on refugee integration, interrupted time series (ITS) analysis is used, with a structural break in time dimension: the sudden change in refugee integration measures after the refugee influx. This method requires collecting multiple data points before and after the intervention; therefore, all possible data points from 2013 to 2020 (eight years) are analyzed. This design controls for any secular trends or temporary fluctuations occurring independently of the intervention. We compared pre-intervention and post-intervention outcome measures such as employment and language adaption to capture the impact of dispersal policy on the economic and social integration of refugees. For instance, our ITS model has the following specification,

yidt = α + β1 Time + B2 Policyt + B3(Time × Policyt ) + Xidt + εidt (2)

y _idt represents the binary indicator of refugee participant i’s employment status, language adaption, living in district d, and interviewed in time t. Where Time is the variable indicating the time period in years. Policy_t is a dummy variable for the post-immigration year that takes the value of 1 if observations are after 2016 and 0 otherwise. The coefficient B₃ estimates the average ITS effect. X_idt is a vector of individual covariates that includes the participant’s age, gender, district-level controls, house prices, and agricultural employment. ɛ_idt is the idiosyncratic error term that captures time-varying unobserved heterogeneity.

5.1 The effect of refugee influx and ethnic clusters on integration

Several factors might create bias in findings. For instance, districts with higher concentrations of a particular ethnic group may have fewer economic resources, such as job opportunities and businesses, making it more difficult for residents to find employment. Another possibility is that discrimination and prejudice based on race or ethnicity may be more prevalent in neighborhoods with higher concentrations of a particular group, making it more difficult for members of that group to access employment opportunities. The complex relationship includes various factors such as economic resources, discrimination, and other social and cultural factors. Therefore, to ensure the DID results are not biased and valid, a series of descriptive analyses across all the treatment and control groups were conducted. These analyses confirmed that treatment and control groups were not significantly different before the refugee influx, and observed differences from the study are solely due to the intervention, i.e., refugee influx rather than preceding differences.

5.1.1 Descriptive statistics

Table 1

Sample characteristics and correlation with output variables
	Observations	Mean	Correlation
			Employment	Language
Refugees (countries)
Arab (Syria, Iraq & Afghanistan)	55,703		-0.16*	-0.07*
Refugees (continents)
Asian refugees	58,889		-0.14*	-0.06*
African refugees	4,322		0.06*	0.02*
European refugees	11,608		0.07*	0.03*
Employment (y/n)	27,333	0.29	-	-
Language adaption(y/n)	25,260	0.72	-	-
Female (gender)	79,285	0.45	-0.26*	-0.16*
Age (years)	78,933	26.79	-0.04*	-0.20*
House prices	-	214.37	0.06*	0.03*
Agricultural employment	-	0.90	-0.085	-0.01*
Note: The table shows descriptive statistics for the sample used in this study

The table above provides a general overview of the study sample, with a total number of observations, the mean of observations across treatment and control groups, and their differences. Lastly, the correlation of the independent variables with the outcome variable portrays the direction and strength of the relationship between our covariates and output variables. The significant correlation for all our variables indicates that each variable has a significant bivariate relationship with the outcome variable, employment, and language adaption (see columns 3 and 4).

Table 2

Treatment and control groups comparison
	Treatment (1)	Control (2)	Difference (3)	Standard error (4)
Arab
Employment	0.51	0.55	0.04	0.03
Oral language adaption	0.88	0.95	0.07	0.02
Asian
Employment	0.52	0.54	0.02	0.03
Oral language adaption	0.88	0.96	0.08	0.02
African
Employment	0.53	0.52	-0.01	0.03
Oral language adaption	0.88	0.95	0.06	0.02
European
Employment	0.52	0.53	0.01	0.03
Oral language adaption	0.89	0.93	0.04	0.02
Note: The average characteristics of the treatment and control and the differences before the refugee influx in 2015 are shown in columns 1 to 3, followed by the Standard Error (S.E) in column 4. The characteristics are shown for all three ethnic groups Arabs (1 to 4) Asians (5 to 8) and Europeans (9 to 12). In parentheses are the robust standard errors. ∗Significant at 10% level; ∗∗significant at 5% level; ∗∗∗significant at 1% level.

The table above shows the mean differences between all the treatment and control groups before the refugee influx. Our results show no significant difference among our measures for economic and social integration, employment, and language adaption, prior to the refugee influx, among the treatment and control districts indicating the differences post-influx are due to the refugee influx and not pre-existing differences. To verify the relationship and impact of refugees being assigned to districts with existing refugee ethnic clusters, employment, and oral language adaption measure is regressed with ethnic refugee share in the general population along with other factors that might influence the outcome variables, such as years of living in the Germany, age, and gender. As the districts might differ in other characteristics, I included district-fixed effects to account for the differences. In table 3, the OLS regression indicates a relationship between ethnic clusters and refugee integration. For instance, the significant negative coefficient for Arab and Asian refugees indicates that ethnic clusters negatively influence employment. However, the analysis is for each year separately from 2013 to 2020; the results indicates no significant association between the two factors before the refugee influx in 2015. There, for our benchmark analysis, we have set the post period from the year of the refugee influx, 2015

Table 3

The relationship between ethnic cluster and refugee integration
	Dependent variable: employment status				Dependent variable: Oral language adaption
	Arab (1)	Asian (2)	African (3)	European (4)	Arab (5)	Asian (6)	African (7)	European (8)
Ethnic refugee share	10.62*** (1.22)	6.83*** (0.98)	42.44*** (14.56)	-15.57 (17.68)	14.81*** (1.20)	12.26*** (1.01)	26.95** (12.29)	-44.12*** (16.81)
Years in Germany	0.03*** (0.001)	0.02*** (0.001)	0.03*** (0.004)	0.01*** (0.002)	0.04*** (0.001)	0.03*** (0.001)	0.05*** (0.003)	0.02*** (0.002)
Age	-0.01*** (0.0002)	-0.01*** (0.0002)	-0.003*** (0.001)	-0.003** (0.001)	-0.01*** (0.0003)	-0.01*** (0.0003)	-0.01*** (0.001)	-0.01*** (0.001)
Gender: Female	-0.24*** (0.01)	-0.25*** (0.01)	-0.30*** (0.02)	-0.13*** (0.03)	-0.16*** (0.01)	-0.16*** (0.01)	-0.20*** (0.02)	-0.06** (0.03)
Observations	17,631	19,214	2,349	1,436	17,019	18,283	2,240	1,097
R-squared	0.1979	0.2006	0.3069	0.3930	0.2043	0.1918	0.2841	0.3736
District fixed effects	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes

Note. This table shows the results the of difference-in-differences method (columns 1 to 8). The observations are at the individual-level unit. The dependent variable measure whether the individual is employed or not. Measured using SOEP questions, are you currently employed? Which one of the following applies best to your status? Dummy variables are created of individuals who voted are employed. In parentheses are the robust standard errors. ∗Significant at 10% level; ∗∗significant at 5% level; ∗∗∗significant at 1% level.

5.1.2 Benchmark analysis

Tables 4 and 5 reports the estimates of equations 1 to 3. In columns (1–8), the generalized DID results show the effect of ethnic clusters on refugee integration due to the 2015-16 refugee influx. Each column and the following column show results of regression with controls. All the regressions are controlled for district-fixed effect. Tables 3 and 4 captures the refugee influx effect on economic and social integration, employment, and German language adaption. The columns represent the results of DID for Arab, Asian, and European samples separately.

The first column shows the estimates of the effect of ethnic clusters on refugee integration without controlling for individual and district-level characteristics. Given that the characteristics mentioned in table 1 likely correlate with employment, we control for all these factors, and the results are shown in the following column (2). The analysis is run across different ethnic groups, Asian refugees (columns 3 and 4) and African refugees (columns 5 and 6), and European refugees (columns 7 and 8). In tables 3 and 4, the estimates compare districts with above median refugee share (treatment) and below median refugee share (control). The estimates from the table show that in treatment districts (districts with higher ethnic share), Arab refugees are 4.60% and 10.63% better at finding employment and adapting to language; however, Asian refugees are 9.27% and 11.43% less likely to find employment and adapt to the German language. Whereas, for African and European refugees, there were no significant differences in employment and language adaption.

As discussed in section 3, the current literature supports both positions regarding the relationship between ethnic clusters and refugee integration. The results of this study contribute to the literature by finding a better understanding of the relationship by illustrating how ethnic background is the decisive factor in the relationship. For instance, (i) the positive relationship between ethnic clusters and integration for war-torn Arab refugees can be related to Marten, Hainmueller, and Hangartner’s (2019) study, where higher ethnic concentration made it easier for refugees to find employment, improving living standards and reduce government’s burden on social benefits. In addition, learning a second language is easier with lower language barriers with the same ethnic teachers. (ii) The negative relationship between ethnic clusters and integration for refugees from the Asian continent is supported by Braun and Dwnger’s (2017) study, where they found segregated neighborhoods to have higher poverty levels and lower economic development, harming both the refugees and the host communities. Relating to social identity and social representation theory, an individual’s behavior or actions would be influenced by their race (i.e., group). For Arab refugees from Syria, Iraq, and Afghanistan, having more group members in a particular district improves their probability of finding jobs or local language adaption compared to areas with lower group members, portraying how collective culture among Asians and Arab works. The membership can be understood to initiate collaboration and teamwork, improving employment opportunities. For instance, a group member would refer another member or help them to find jobs. These results align with Marten, Hainmueller, and Hangartner (2019) and Fasani, Frattini, and Minale (2021). Whereas, in comparison, the negative influence of the Asian ethnic cluster can be related to several factors. First, refugees with limited German language proficiency might suffer in communicating with people outside of the ethnic cluster due to lack of interaction with the host population, harming both host language adaption capabilities, and job opportunities. Second, these ethnic enclaves inadvertently block exposure to local culture and community, harming further interaction and, therefore, successful integration. Third, the Asian refugee sample for this study has refugees from different geographical and cultural backgrounds, ranging from South Asian countries such as Bangladesh, and Pakistan to Eastern Asian countries such as the Philippines and Vietnam to Central Asian countries such as Azerbaijan and Uzbekistan. Therefore, relating to Tajfel and Turner’s (1979) formation of social groups helping to integrate might be comparatively more difficult.

On the contrary, African and European ethnic clusters showed no influence in finding employment and adapting local language. Though Africans are collectivist in nature, the hundreds of language dialects across the population might offset the influence on the formation of ‘groups’ based on language or ethnicity, which might influence integration measures. However, several other reasons might influence this relationship which the model is unable to capture, such as perception, attitude, and eagerness to integrate. The probable reason behind ethnic clusters having no influence on refugee integration for Europeans might be due to two major reasons. First, in general, Europeans are understood to be individualistic in nature, compared to the collectivist nature of Asians. Individualism in nature favors freedom of action or not being dependent like collectivist groups. Second, Europeans find it hard to differentiate and create smaller groups, as the majority are Europeans or ‘white’ in Germany’s content. Comparatively, minor groups find it easier to organize by virtue of their small size and tend to focus on a narrow set of issues (Saideman, 2022). Nonetheless, as significant literature and policymakers have emphasized ethnic clusters to negatively harm integration, I further test the findings across districts that are outliers, in terms of receiving the highest number of refugees in table 6.

Table 4

The effect of ethnic clusters on refugee employment
	Arabs (1)	+controls (2)	Asians (3)	+controls (4)	Africans (5)	+controls (6)	Europeans (7)	+controls (8)
Ethnic refugee numbers	8.50*** (3.44)	7.87*** (2.84)	30.65*** (0.11)	20.27*** (3.80)	98.33*** (29.16)	87.48*** (27.45)	-4.05 (21.22)	-11.78 (21.86)
Post influx	-0.44*** (0.03)	-0.13*** (0.04)	-0.24*** (0.03)	-0.12*** (0.03)	-0.41*** (0.07)	-0.20*** (0.073)	-0.06 (0.04)	-0.11** (0.05)
Ethnic refugee numbers * post influx	13.85*** (2.63)	4.60** (2.24)	-17.27*** (3.61)	-9.27*** (3.35)	-9.05 (16.39)	-16.06 (15.72)	-2.52 (9.71)	7.76 (10.36)
Years in Germany		0.03*** (0.001)		0.02*** (0.001)		0.03*** (0.004)		0.01*** (0.002)
Age		-0.01*** (0.0003)		-0.01*** (0.0002)		-0.003*** (0.001)		-0.003** (0.001)
Gender: Female		-0.24*** (0.01)		-0.25*** (0.01)		-0.30*** (0.02)		-0.13*** (0.03)
Observations	17,077	17,631	19,287	19,214	2,364	2,349	1,624	1,436
R-squared	0.0682	0.1987	0.0769	0.2026	0.2042	0.3096	0.3666	0.3962
District fixed effects	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes

Note. This table shows results of difference-in-differences method (columns 1 to 6). The observations are at the individual-level unit. The dependent variable measure whether the individual is employed or not. Measured using SOEP questions, are you currently employed? Which one of the following applies best to your status? Dummy variables are created of individuals who voted are employed. In parentheses are the robust standard errors. ∗Significant at 10% level; ∗∗significant at 5% level; ∗∗∗significant at 1% level.

Table 4

The effect of ethnic clusters on refugee German language adaption
	Arabs (1)	+controls (2)	Asians (3)	+controls (4)	Africans (5)	+controls (6)	Europeans (7)	+controls (8)
Ethnic refugee numbers	8.50** (3.44)	4.03 (3.09)	40.95*** (4.16)	27.28*** (3.89)	51.23** (27.58)	28.70 (24.96)	-44.04* (24.51)	-38.60* (23.15)
Post influx	-0.44*** (0.03)	-0.07** (0.04)	-0.25*** (0.03)	-0.09*** (0.03)	-0.37*** (0.07)	-0.06 (0.06)	-0.01 (0.05)	-0.03 (0.05)
Ethnic refugee numbers * post influx	13.85*** (2.63)	10.63*** (2.43)	-21.83*** (3.47)	-11.43*** (3.34)	12.71 (15.24)	2.87 (14.48)	-1.96 (11.52)	-0.54 (10.90)
Yeas in Germany		0.04*** (0.001)		0.03*** (0.001)		0.05*** (0.004)		0.02*** (0.002)
Age		-0.01*** (0.0003)		-0.01*** (0.0003)		-0.01*** (0.001)		-0.01*** (0.001)
Gender: Female		-0.16*** (0.01)		-0.16*** (0.01)		-0.20*** (0.02)		-0.06** (0.03)
Observations	17,077	17,019	18,345	18,283	2,252	2,240	1,117	1,097
R-squared	0.0682	0.2053	0.0641	0.1932	0.1693	0.2843	0.2851	0.3740
District fixed effects	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes

Note. This table shows results of difference-in-differences method (columns 1 to 6). The observations are at the individual-level unit. The dependent variable measures whether the individual can speak German. Measured using SOEP questions, how well can you speak German? Dummy variables are created of individuals who can speak very good and good. In parentheses are the robust standard errors. ∗Significant at 10% level; ∗∗significant at 5% level; ∗∗∗significant at 1% level

Table 6

The effect of ethnic clusters on refugee employment and language adaption across districts with largest refugee numbers
	Arab		Asia
	Employment (1)	Oral language (2)	Employment (3)	Oral language (4)
Ethnic refugee numbers	14.74*** (1.41)	17.52*** (1.49)	11.56*** (1.22)	16.73*** (1.26)
Post influx	-0.14*** (0.04)	-0.08*** (0.04)	-0.16*** (0.03)	-0.15*** (0.03)
Ethnic refugee numbers * post influx	-6.42*** (2.36)	-6.55*** (1.95)	-2.33*** (1.01)	-3.51*** (1.03)
Yeas in Germany	0.03*** (0.001)	0.04*** (0.001)	0.02*** (0.001)	0.03*** (0.001)
Age	-0.01*** (0.0003)	-0.01*** (0.0003)	-0.01*** (0.0002)	-0.01*** (0.0003)
Gender: Female	-0.24*** (0.01)	-0.16*** (0.01)	-0.25*** (0.01)	-0.16*** (0.01)
Observations	17,631	17,019	19,214	18,283
R-squared	0.1991	0.2049	0.2024	0.1932
District fixed effects	Yes	Yes	Yes	Yes

Note. This table shows the results of difference-in-differences method for varying ethnic intensity for Arabs (columns 1 to 3), Asians (columns 4 to 6), and European (columns 7 to 9). The observations are at the individual-level unit. The dependent variable measures whether individuals is employed or not. Measured using SOEP questions, are you currently employed? Which one of the following applies best to your status? Dummy variables are created of individuals who voted are employed. In parentheses are the robust standard errors. ∗Significant at 10% level; ∗∗significant at 5% level; ∗∗∗significant at 1% level.

From the table above, the results are found to be consistent with the benchmark analysis. However, for Arab refugees, areas with the highest percentile of ethnic refugee share influence employment and language adaption, measures for economic and social integration, in the opposite direction; negatively. This alarms how the size of the refugee share allocated in districts plays an important role in the integration process.

The size of refugee ethnic share influences the significance and direction of the relationship between ethnic clusters and integration for Arab refugees fleeing war (see columns 1 and 2 of table 5). For instance, when comparing districts with refugee share above the median, the relationship is positive; however, when comparing districts with only the highest refugee share, it has a statistically significant negative effect. This finding justifies the dispersal policies adopted by governments across European nations to tackle ethnic segregation to promote the successful integration of refugees. Nevertheless, the difference in findings across different ethnic backgrounds raises the question about the effectiveness of one for-all policy, for example, the dispersal policy of the Integration Act 2016, restricting the movement of refugees from all ethnic backgrounds in Germany to avoid ethnic segregation. Rather, tailored policies for groups of different demographic, backgrounds, and ethnicity catered to specific characteristics might be more fruitful, as groups might guide individuals’ behavior and self-knowledge (Tajfel and Turner, 1979).

5.1.4 Robustness checks

5.1.4.1 Propensity score matching difference in differences (DID)

Though the allocation of refugees across districts is considered random, as refugees cannot self-select where they will live, the allocation decision might be influenced by various integration factors, such as the availability of housing and support services. Therefore, I used propensity score matching technique to control for the effect of house prices on the assignment of refugees across districts. Propensity scores allow us to estimate the prospect of being assigned to a treatment district based on house price, creating similar treatment and control pairs with similar propensity scores. Propensity score matching validates whether comparison across groups is not biased, functioning as a robustness check of our benchmark results under different assumptions and conditions, justifying whether they are robust under different model specifications.

The common support estimation is used, which mentions the range of values of covariates in overlapping treatment and control groups to ensure a valid comparison between the groups. The region of common support (see row 1) represents the range of values shared among treatment and control groups, providing a more accurate estimate of the treatment effect of the refugee influx. For instance, by limiting the analysis to regions of common support, the estimation is expected to control for potential differences between the groups, providing a more accurate treatment effect estimation. The propensity score mean indicates a higher probability of individuals getting treatment. Lastly, as the covariates show balanced properties, it indicates this analysis is successfull in achieving similar distribution of covariates between treatment and control groups after matching, ensuring the differences in our output variable are due to the refugee influx rather than pre-existing differences.

Table 7

Characteristics of propensity score matching
	Arab	Asian	African	European
Region of common support	[0.70, 0.98]	[0.31, 0.99]	[0.34,0.99]	[0.28,0.99]
Propensity score means	0.77	0.88	0.63	0.59
Standard deviation	0.03	0.06	0.14	0.11
Balance on covariate	Yes	Yes	Yes	Yes
Observations	18,053	19,680	2,361	1,437

Note: This table shows the charactertistics of propensity score matching, Arabs (columns 1 to 3) and Asians (columns 4 to 6). The observations are at the individual-level unit. The dependent variable measures whether individuals is employed or not and whether the individual can speak German.

Lastly, I have used Nearest Neighbor Matching method while bootstrapping standard errors to balance the potential imbalance between the treatment and control groups (see table 8).

Table 8

The average treatment effects using the nearest neighbor matching method
	n Treatment	N Control	ATT	Std error	t value
Arab
Employment	13888	3476	0.027	0.006	4.220
Language adaption	13888	3476	0.079	0.006	12.478
Asia
Employment	17287	2147	0.067	0.011	5.841
Language adaption	17287	2147	0.105	0.021	5.029
Africa
Employment	1500	542	0.159	0.028	5.592
Language adaption	1500	542	0.105	0.047	2.242
Europeans
Employment	846	362	0.482	0.029	16.623
Language adaption	846	362	0.541	0.028	19.328

Note. This table shows results of average treatment effects using nearest neighbor matching. The observations are at the individual-level unit. The dependent variable measures whether individuals is employed or not, and whether the individual can speak German language.

The table above shows the average treatment effects on treated (ATT) estimation using the nearest neighbor matching method (see column 3). The significant t-values indicate that the difference in output variables, employment status, and German language adaption is different between the treatment and control group after matching for covariates. This proves that the refugee influx had a significant effect on the treatment group, further proving the robustness of our model.

5.1.4.2 Falsification test

To test the robustness of the model, two hypothetical scenarios are established, (i) the treatment group is manipulated to contain districts with low refugee share (ii) where the post-influx period is considered as the prior years rather than post-years. I conducted the analysis across Arab and Asian refugee groups. From the results in table 9, it can be observed that the coefficients remain insignificant, and therefore believe that the data supports the assumptions and operations in the model used. The falsification test further proves that the effect on the integration measures is solely due to the refugee influx rather than any other simultaneously occurring factors or events.

Table 9

Falsification tests
	False treatment area – low refugee share and higher				False influx year – 2013 and 2014
	Arab		Asia		Arab		Asia
	Employment status (1)	Oral language (2)	Employment status (3)	Oral language (4)	Employment Status (5)	Oral language (6)	Employment status (7)	Oral language (8)
Ethnic refugee numbers	12.71** (5.0)	23.44*** (4.94)	6.71 (5.50)	6.71** (5.50)	12.98*** (1.38)	15.88*** (1.41)	10.30*** (1.18)	14.97*** (1.21)
Post influx	-0.13*** (0.04)	-0.04 (0.04)	-0.16*** (0.04)	-0.16*** (0.04)	0.13*** (0.04)	0.06*** (0.04)	0.14*** (0.03)	0.13*** (0.03)
Ethnic refugee numbers * post influx	0.26 (4.62)	-7.20 (4.45)	3.49 (5.19)	3.49 (5.19)	-7.54 (15.69)	-7.52 (15.83)	3.34 (7.03)	0.92 (5.78)
Yeas in Germany	0.03*** (0.001)	0.04*** (0.002)	0.02*** (0.001)	0.02*** (0.001)	0.03*** (0.001)	0.04*** (0.001)	0.02*** (0.001)	0.03*** (0.001)
Age	-0.01*** (0.0002)	-0.01*** (0.0003)	-0.01*** (0.0002)	-0.01*** 0.0002)	-0.005*** (0.0003)	-0.01*** (0.0003)	-0.01*** (0.0002)	-0.01*** (0.0003)
Gender: Female	-0.24*** (0.01)	-0.16*** (0.01)	-0.23*** (0.01)	-0.25*** (0.01)	-0.24*** (0.01)	-0.16*** (0.01)	-0.25*** (0.01)	-0.16*** (0.01)
Observations	17,631	17,019	19,214	19,214	17,631	17,019	19,214	18,283
R-squared	0.1985	0.2045	0.2022	0.2022	0.1985	17,019	0.2022	0.1927
District fixed effects	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes

Note. This table shows results of the difference-in-differences method for falsification tests for Arabs (columns 1,2,5 and 6) and Asians (columns 3,4,7 and 8). The observations are at the individual-level unit. The dependent variable measures whether individuals is employed or not and whether the individual can speak German language. Measured using SOEP questions, are you currently employed? Which one of the following applies best to your status? Dummy variables are created of individuals who voted are employed. In parentheses are the robust standard errors. ∗Significant at 10% level; ∗∗significant at 5% level; ∗∗∗significant at 1% level.

5.1.4.3 Sensitivity test

In this study, a difference in difference (DID) analysis has been used to investigate the influence of ethnic clusters on refugee integration. Nevertheless, due to the sample being limited to only refugees of the SOEP data, the results might be subjected to selection bias. For instance, a significant portion of refugees might be considered as migrants or migrant in the sample might have been naturalized from refugees in the past. Referring to SOEP data, migrants in late 1900s are considered as migrants, however, thousands fled the Yugoslavian wars to Germany during that period, therefore considered to be refugees. Therefore, along with the refugee sample, we include the entire migrant sample.

Table 10

Sensitivity test using larger sample
	Arabs		Asian
	Employment status (1)	Oral language (2)	Employment status (3)	Oral language (4)
Ethnic refugee numbers	7.47*** (2.82)	3.83 (3.03)	19.76*** (2.80)	23.63*** (2.51)
Post influx	-0.13*** (0.04)	-0.09*** (0.03)	-0.12*** (0.02)	-0.15*** (0.01)
Ethnic refugee numbers * post influx	4.66* (2.21)	10.40*** (2.37)	-15.69*** (2.30)	-12.89*** (1.97)
Yeas in Germany	0.03*** (0.001)	0.04*** (0.001)	0.02*** (0.0004)	0.02*** (0.0004)
Age	-0.01*** (0.0002)	-0.01*** (0.0003)	-0.01*** (0.0002)	-0.01*** (0.0002)
Gender: Female	-0.24*** (0.01)	-0.16*** (0.01)	-0.21*** (0.005)	-0.14*** (0.01)
Observations	17,853	17,157	24,744	21,723
R-squared	0.2001	0.2044	0.2275	0.1906
District fixed effects	Yes	Yes	Yes	Yes

Note. This table shows results of the difference-in-differences method for placebo test using random allocation for Arabs (columns 1 and 3), Asians (columns 1 and 3) and Europeans (columns 3 and 6). The observations are at the individual-level unit. The dependent variable measures whether individuals is employed or not and whether the individual can speak German language. Measured using SOEP questions, are you currently employed? Which one of the following applies best to your status? Dummy variables are created of individuals who voted are employed. In parentheses are the robust standard errors. ∗Significant at 10% level; ∗∗significant at 5% level; ∗∗∗significant at 1% level.

The results show the interaction coefficients remain significant and consistent with the benchmark analysis for Arabs, Asians and European samples. Therefore, indicating further robustness of the model used in this study.

5.1.4.4 Placebo test

To further test the robustness of the model, participants from the sample are randomly allocated to either the treatment or control group. The model is run with the new placebo treatment and control groups. As placebos are inactive treatment and control, insignificant results will further indicate the robustness of our model specifications.

Table 11

Placebo test using random allocation
	Arabs		Asian
	Employment status (1)	Oral language (2)	Employment status (3)	Oral language (4)
Ethnic refugee numbers	-0.88 (14.37)	-7.03 (14.04)	-2.43*** (9.60)	1.54 (8.46)
Post influx	-0.23*** (0.06)	-0.13*** (0.06)	-0.22** (0.05)	-0.22*** (0.05)
Ethnic refugee numbers * post influx	12.03 (14.00)	7.09 (13.79)	11.01 (9.39)	11.59 (8.20)
Age	0.03*** (0.00)	0.04*** (0.002)	0.02*** (0.001)	0.02*** (0.001)
Gender: Female	-0.01*** (0.0004)	-0.01*** (0.0004)	-0.01 *** (0.0004)	-0.01*** (0.0004)
House price	-0.23*** (0.01)	-0.17*** (0.01)	-0.25*** (0.01)	-0.16*** (0.01)
Agricultural employment
Observations	9,048	8,501	10,059	9,198
R-squared	0.2294	0.2191	0.2388	0.2049

Note. This table shows results of the difference-in-differences method for placebo test using random allocation for Arabs (columns 1 and 3) and Asians (columns 1 and 3). The observations are at the individual-level unit. The dependent variable measures whether individuals is employed or not and whether the individual can speak German language. Measured using SOEP questions, are you currently employed? Which one of the following applies best to your status? Dummy variables are created of individuals who voted are employed. In parentheses are the robust standard errors. ∗Significant at 10% level; ∗∗significant at 5% level; ∗∗∗significant at 1% level.

The insignificant coefficient for the DID interaction terms supports the robustness of the model. The refugee influx has no significant effect on placebo treatment and control groups, which indicates that the assumptions and model specification is effective.

5.2 The effect of the dispersal policy of the Integration Act 2015

The Integration Act 2016 was passed in August of 2016 and aimed to improve the integration of immigrants and refugees into German society. Though the Interaction Act introduced other measures such as language classes, education, training, and work visa scope and extension, such measures benefit all refugees, only the movement restriction is applied to refugees who migrated in the year 2016 and onwards. To investigate the effectiveness of the dispersal policy in the Integration Act 2016, which restricts refugee movement from their assigned location for refugees who immigrated in the year 2016 and onwards, I conducted an Interrupted Time series (ITS) analysis. The implementation of the ITS estimator builds on an assumption, excluding structural change, past events are the most effective predictors of future occurrences (Bloom, 2003). Therefore, the model requires that without the change of interest, the policy implementations, the outcome variables would have been equivalent to an extrapolation based on previous information available.

5.2.1 Descriptive statistics

The figure above portrays a sudden change in the mean level and trend of the time series for refugees who immigrated in the year 2016 and onwards, implying the policy intervention had a significant effect on employment and language adaption; economic and social integration of refugees. Therefore, a t-test analysis was used to verify these visual findings.

Table 12

The mean difference and standard error pre and post-policy intervention
	Post policy (1)	Pre policy (2)	Difference (3)	Standard error (4)
Employment status	0.16	0.31	-0.15***	0.01
German language	0.61	0.74	-0.13***	0.01
Age	26.09	26.91	0.82	0.17
Gender: Female	0.44	0.49	-0.05*	0.01
House prices	221.50	213.14	-8.36***	5.39
Agricultural employment	0.93	0.90	0.04	0.12

Note. This table shows the mean difference (column 3) and standard error (column 4) before and after the policy intervention.

The t-test results further confirm that economic and social integration has significantly differed between post and pre-policy implementation. For instance, our outcome variables, employment status, and language adaption show significant differences between the two groups (see column 3).

In the context of the ITS approach, there are several threats to the identification. First, the existence of other structural or temporary changes that might influence integration measures (Baicker and Svoronos, 2019). For instance, as ITS estimates capture the changes for a particular period, it cannot extract the effect of the particular effect interest from other simultaneous interventions. The second threat is predicting the change, implying that pre-intervention information used to fit the model without a break would be biased. The settings of our study help to prevent these threats. There are no other factors, such as economic change or social hardships, that might influence the employment or language adaption differences across immigrants immigrating before and after the year 2016. For instance, the unemployment rates that would capture economic changes or employment disruption decreased over the majority of the period of analysis from 2013 to 2019 (International Labour Organization, 2022), and language courses applied to refugees regardless of immigration year. Second, the movement restriction from the policy can be considered unpredictable as the restriction applied to refugees who arrived in Germany even before the Integration Act was passed. For instance, the Integration Act 2016 was passed in August of 2016, which adopted the law of movement restriction for refugees who arrived in Germany in January 2016. In addition, refugees fleeing war and seeking asylum are less likely to be familiar with foreign governments’ future plans or policies.

5.2.2 Benchmark analysis

Table 13

The effect of dispersal policy on refugee integration
	Employment (1)	+controls (2)	Language adaption (3)	+controls (4)
Year	0.003** (0.002)	-0.01*** (0.002)	-0.01*** (0.002)	-0.02*** (0.002)
Immigration year 16 on	50.03*** (6.22)	-51.15*** (8.47)	-95.11*** (8.85)	-92.42*** (8.74)
Year * immigration year 16	0.02*** (0.003)	0.03*** (0.004)	0.05*** (0.004)	0.05*** (0.004)
Age		-0.004*** (0.0002)		-0.01*** (0.0002)
Gender: Female		-0.24 *** (0.001)		-0.15*** (0.01)
House price		-4.31e-06*** (0.00002)		0.00002 (0.00002)
Agricultural employment		-0.04*** (0.03)		-0.06 (0.04)
Observations	27,331	25,979	26,971	23,973
R-squared	0.14	0.21	0.14	0.21
District fixed effects	Yes	Yes	Yes	Yes
Time fixed effects	Yes	Yes	Yes	Yes

Note. This table shows results of ITS regression analysis. The observations are at the individual-level unit. The dependent variable measures whether individuals is employed or not and whether the individual can speak German language. Measured using SOEP questions, are you currently employed? Which one of the following applies best to your status? Dummy variables are created of individuals who voted are employed. In parentheses are the robust standard errors. ∗Significant at 10% level; ∗∗significant at 5% level; ∗∗∗significant at 1% level.

From the table above, the results indicate that refugees who migrated to Germany in 2016 and afterward and had movement restrictions for a minimum of two years were slightly less likely to find employment and adapt to the German language than refugees who migrated before 2016. Due to movement restrictions, refugees might be seized from having access to social networks, information, and referrals about job opportunities within their community. Refugees allowed to relocate to areas with a shared language and cultural background would mean a better possibility of finding employment. The individuals might be considered more reliable and trustworthy compared to a less diverse neighborhood, facing discrimination. Lastly, a possibility of a mismatch between skills and local labor market demand might remain, making finding employment worse. The harm in German language adaption can be related to limited access to special language classes, stress and anxiety from alienation, and other factors. Though language classes are being made accessible, not all might be highly effective. For instance, some language centers might be better at helping refugees than others; however, movement restriction makes them inaccessible. Language adaption might be enhanced through interaction with the general population, but not all populations across districts are open and friendly towards refugees, harming interaction. In addition, the restriction to relocate to a more friendly environment or friends and family further harms refugees through stress and anxiety. Lastly, the best tutors or language exchange partners of a foreign language might be individuals from their own ethnic community, gaining extra support and guidance while learning. This can provide a more familiar and comfortable environment while learning a new language. Similar findings were found by Fasani, Frattini, and Minale (2021). They analyzed the efficacy of dispersal policies across Denmark, Ireland, Netherlands, Norway, and the UK by conducting a difference-in-difference analysis. Their quasi-experimental study showed that such policies worsened the economic integration of refugees as labor market participation suffered. Furthermore, employment bans and restrictions harmed refugee integration prospects for approximately ten years after arrival (Fasani, Frattini, and Minale, 2022).

Nevertheless, the positive significant interaction term (row 3 table of 12) indicates that the influence of policy on outcome variables, employment, and language adaption, changed positively over time. For instance, the policy initially had a negative effect on employment and language adaption, which contracted with time. There are a few probable reasons behind this phenomenon. First, the policy restricts movement temporarily, ranging from two years onwards. Refugees are free to move to new locations afterward. Second, the movement restriction might be lifted if an individual or their immediate family finds job opportunities in other districts. Third, with time refugees are more likely to interact and make social connections, improving employment and language adaption. Fourth, refugees might opt for online job portals and language courses to find employment and improve local language proficiency. The findings that movement restrictions harmed economic and social integration in the short run but later improved with time relate to the social mismatch theory by Kain (1968), emphasizing the importance of social connections and networks (Fried, 1999). For instance, Kling, Liebman and Katz (2005) found that a mismatch between low-skilled workers and jobs might lead to negative employment outcomes. Similarly, Lambert & Tucker (1972) have found that interaction and exposure to local culture and language improve second language adaption. The gradual improvement in integration with time might be the acculturation process, where migrants and refugees require time to blend into the new culture while maintaining their own identity (Kuhlman, 1991).

5.2.3 Robustness checks

To test the robustness of our model, we conducted two validity tests. First, the analysis was run for refugee samples who had immigrated to Germany before 2016 and had no movement restrictions. Second, refugees, regardless of the immigrant year, were randomly allocated to the treatment group, which faced movement restriction.

Table 14

falsification test
	Employment status (1)	Language adaption (2)
Year	-0.02*** (0.003)	-0.03*** (0.003)
Immigration year	-25.93*** (9.56)	65.23*** (6.05)
Year * immigration year	0.01*** (0.005)	0.03*** (0.01)
Age	-0.003*** (0.0001)	-0.01*** (0.0002)
Gender: Female	-0.24*** (0.01)	-0.15*** (0.01)
House price	4.11e-06 (0.00003)	0.00002 (0.00002)
Agricultural employment	-0.06 (0.05)	-0.07 (0.04)
Observations	25,979	23,973
R-squared	0.19	0.20

Note. This table shows results of ITS regression analysis for falsification tests. The observations are at the individual-level unit. The dependent variable measures whether individuals is employed or not and whether the individual can speak German language. Measured using SOEP questions, are you currently employed? Which one of the following applies best to your status? Dummy variables are created of individuals who voted are employed. In parentheses are the robust standard errors. ∗Significant at 10% level; ∗∗significant at 5% level; ∗∗∗significant at 1% level.

The results indicated that refugees who immigrated before 2016 have a significant positive effect on employment and language adaptation, proving the robustness of our model. Second, as expected, there was no significant effect of policy on integration measures. The data source, SOEP, is a nationally representative survey that used sampling techniques formed on characteristics like gender, age, and other factors to establish our experimental groups. Therefore, there is a probability of the sample missing out on individuals who might be influenced by the policy, compromising the validity of our results. Therefore, to mitigate this, two random subsets from our sample were used to examine whether results remain consistent across sub-samples.

Table 15

External validity test – random allocation
	Employment status (1)	Language adaption (2)
Year	0.01*** (0.003)	-0.01*** (0.002)
Immigration year	0.10*** (0.01)	-0.15*** (0.01)
Year * immigration year	0.12*** (0.01)	0.07*** (0.01)
Age	-0.03*** (0.0002)	-0.01*** (0.0002)
Gender: Female	-0.23*** (0.01)	-0.12*** (0.01)
House price	-5.32e-06 (0.00002)	0.00002 (0.00002)
Agricultural employment	-0.04 (0.04)	-0.06* (0.04)
Observations	25,979	23,973
R-squared	0.20	0.20

Note. This table shows results of ITS regression analysis for external validity. The observations are at the individual-level unit. The dependent variable measures whether individuals is employed or not and whether the individual can speak German language. Measured using SOEP questions, are you currently employed? Which one of the following applies best to your status? Dummy variables are created of individuals who voted are employed. In parentheses are the robust standard errors. ∗Significant at 10% level; ∗∗significant at 5% level; ∗∗∗significant at 1% level.

The results were the same across the subsets, and the interaction term coefficients were notably significant, lending credibility to the external validity of the findings.

The findings of this study have implications for both current literature and policy-makers. From an academic perspective, the mixed results for integration measures across different races might initiate the need for further research to understand the factors behind this phenomenon. For policymakers, the study highlights the importance of examining specific needs and challenges for refugees from different backgrounds while designing integration policies. Similarly, segregation policies might influence refugees from different ethnic backgrounds, and policymakers should consider targeted measures for the specific needs and challenges of different ethnic refugees to promote successful integration into the local community.

In conclusion, the findings show that the influence of ethnic clusters on refugee integration differs across different ethnic backgrounds. And the dispersal policy of the Integration Act 2016 has a short-term negative effect on integration measures, which improves and gets positive with time. For this study, the data used are from secondary data resources, SOEP, and the German statistical office to conduct DID and ITS analysis to investigate the effect of the influx-driven ethnic cluster and the dispersal policy on refugee integration. I compared treatment districts, districts with greater than the average median refugee share among districts, with control districts, districts below the average median refugee share. The results show that Asian refugees in districts with greater ethnic share (treatment) tend to have lower employment rates and language adaption than districts with fewer ethnic refugees (control). However, refugees from war-torn Arab countries, such as Syria, Iraq, and Afghanistan, are better at finding employment and oral language adaption in districts with higher ethnic share than districts with lower ethnic share. However, when I compared integration measures in districts with very high refugee share, the positive relationship turned negative, as the disadvantages overplay the advantages of collective benefits if the refugee share is too high. Lastly, the results showed no significant effects across African and European refugees’ integration measures. The findings can be related to the major language differences among Africans and the individualistic nature of Europeans.

The ITS analysis results show dispersal policy of the Integration Act 2016 negatively affected the economic and social integration of refugees, which became positive with time. As discussed in social identity theory, these findings shed light on how individuals associate or conceptualize with their groups. The positive effects of ethnic clusters on Arab and Asian refugees illustrate how ethnic clusters or groups can build group identity, benefiting the well-being and integration of refugees. Furthermore, the dispersal policy to avoid ethnic clusters from forming might have disrupted the sense of group identity, harming integration measures. However, the improvement over time indicates that refugees might have developed new associations or identities along with the new experience in their new environment. The gradual phenomenon of integrating getting better with time can be related to Kuhlman’s (1991) concept of integration, where he discusses how acculturation is a process requiring refugees to learn and adapt to the new cultural values and beliefs, building a new sense of identity without harming their unique differences. Though the results are robust across several falsification and placebo tests, the study has certain limitations. For instance, the study was conducted at the district level, and a more micro-level understanding, such as a study across municipalities, might provide better insights. Additionally, in-depth knowledge through qualitative research, such as in-depth interviews and focus studies, can provide a more holistic understanding of the relationship between ethnic clusters and refugee integration. Overall, the mixed results across different ethnic groups of refugees support social identity theory’s assumption that group identities and perceptions of social status can influence integration experiences. Thus, policymakers can study factors behind such differences in integration to develop strategies and promote better positive integration outcomes for refugees.

Competing interests: The authors declare no competing interests.

Ager, A., Strang, A., & Wessells, M. (2006). Integrating psychosocial issues in humanitarian and development assistance: a response to Williamson and Robinson. Intervention, 4(1), 29-31.
Ager, A. and Strang, A., 2008. Understanding integration: A conceptual framework. Journal of refugee studies, 21(2), pp.166-191.
Augoustinos, M., 2001. Social categorization: Towards theoretical integration. Blackwell Publishers Inc.
Baicker, K. and Svoronos, T., 2019. Testing the validity of the single interrupted time series design (No. w26080). National Bureau of Economic Research.
Battisti, M., Peri, G. and Romiti, A., 2016. Dynamic effects of Co-Ethnic networks on immigrants' economic success (No. w22389). National Bureau of Economic Research.
Bekaj, A. and Antara, L., 2018. Political participation of refugees: Bridging the gaps.
Bloom, H.S., 2003. Using “short” interrupted time-series analysis to measure the impacts of whole-school reforms: With applications to a study of accelerated schools. Evaluation Review, 27(1), pp.3-49.
BAMF, 2016. Das Bundesamt in Zahlen 2015—Modul Asyl. Technical Report, Federal Office for Migration and Refugees (BAMF).
Borjas, G.J. and Monras, J., 2016. The labor market consequences of refugee supply shocks (No. w22656). National Bureau of Economic Research.
Borselli, M. and van Meijl, T., 2021. Linking migration aspirations to integration prospects: The experience of Syrian refugees in Sweden. Journal of Refugee Studies, 34(1), pp.579-595.
Braun, S. and Dwenger, N., 2017. The local environment shapes refugee integration: Evidence from post-war Germany (No. 10-2017). Hohenheim Discussion Papers in Business, Economics and Social Sciences
Christ, S., 2020. The key to successful integration of refugees in Germany: language skills, housing, work, social participation.
Damm, A.P., 2009. Determinants of recent immigrants’ location choices: quasi-experimental evidence. Journal of population Economics, 22(1), pp.145-174.
De Haas, H., 2008. The myth of invasion: The inconvenient realities of African migration to Europe. Third world quarterly, 29(7), pp.1305-1322.
Degler, E. and Liebig, T., 2017. Finding Their Way. Labour Market Integration of Refugees in Germany. OECD.
Del Carpio, X.V. and Wagner, M.C., 2015. The impact of Syrian refugees on the Turkish labor market. World Bank policy research working paper, (7402).
Dustmann, C., Glitz, A., Schönberg, U. and Brücker, H., 2016. Referral-based job search networks. The Review of Economic Studies, 83(2), pp.514-546.
Edin, P.A., Fredriksson, P. and Åslund, O., 2003. Ethnic enclaves and the economic success of immigrants—Evidence from a natural experiment. The quarterly journal of economics, 118(1), pp.329-357.
Fasani, F., Frattini, T. and Minale, L., 2021. Lift the ban? Initial employment restrictions and refugee labour market outcomes. Journal of the European Economic Association, 19(5), pp.2803-2854.
Fasani, F., Frattini, T. and Minale, L., 2022. (The Struggle for) Refugee integration into the labour market: evidence from Europe. Journal of Economic Geography, 22(2), pp.351-393.
Fontanari, E., 2022. Germany, year 2020. The tension between asylum right, border control, and economy, through the imperative of deservingness. Migration Studies.
Gehrsitz, M. and Ungerer, M., 2022. Jobs, Crime and Votes: A Short‐run Evaluation of the Refugee Crisis in Germany. Economica, 89(355), pp.592-626.
Gordon, M.M., 1964. Assimilation in American life: The role of race, religion, and national origins. Oxford University Press on Demand.
Hajak, V.L., Sardana, S., Verdeli, H. and Grimm, S., 2021. A systematic review of factors affecting mental health and well-being of asylum seekers and refugees in Germany. Frontiers in psychiatry, 12, p.643704.
Hartnett, A.S., 2018. The effect of refugee integration on migrant labor in Jordan. Review of Middle East Studies, 52(2), pp.263-282.
Hogg, M.A., 2013. Intergroup relations. In Handbook of social psychology (pp. 533-561). Springer, Dordrecht.
Hynie, M., Korn, A., & Tao, D. (2016). Social context and social integration for Government Assisted Refugees in Ontario, Canada. After the flight: The dynamics of refugee settlement and integration, 183-227.
International Labour Organization. (2022). iLOSTAT. Available at: https://ilostat.ilo.org/data/ [Accessed: 07 February 2023].
Jackson, S. and Bauder, H., 2014. Neither temporary, nor permanent: The precarious employment experiences of refugee claimants in Canada. Journal of Refugee Studies, 27(3), pp.360-381.
Jargowsky, P.A., 2002. Sprawl, concentration of poverty, and urban inequality. Urban sprawl: Causes, consequences, and policy responses, pp.39-72.
Kain, J.F., 1968. Housing segregation, negro employment, and metropolitan decentralization. The quarterly journal of economics, 82(2), pp.175-197.
Kalter, F. and Kogan, I., 2014. Migrant networks and labor market integration of immigrants from the former Soviet Union in Germany. Social Forces, 92(4), pp.1435-1456.
Kling, J.R., Liebman, J.B. and Katz, L.F., 2005. Bullets don’t got no name: Consequences of fear in the ghetto. Discovering successful pathways in children’s development: Mixed methods in the study of childhood and family life, pp.243-281.
Kogan, I., 2011. New immigrants―old disadvantage patterns? Labour market integration of recent immigrants into Germany. International Migration, 49(1), pp.91-117.
Kristiansen, M.H., Maas, I., Boschman, S. and Vrooman, J.C., 2022. Refugees’ transition from welfare to work: A quasi-experimental approach of the impact of the neighbourhood context. European Sociological Review, 38(2), pp.234-251.
Kristjansdottir, E.S. and Christiansen, T., 2017. "... you have to face the fact that you're a foreigner": Immigrants' Lived Experience of Communication and Negotiation Position Toward their Employer in Iceland. Journal of Intercultural Communication, (44), pp.N_A-N_A.
Kuhlman, T. (1991). The economic integration of refugees in developing countries: a research model. Journal of refugee studies, 4(1), 1-20.
Lambert, W.E. and Tucker, G.R., 1972. Bilingual education of children: The St. Lambert experiment.
Landau, L.B., 2006. Protection and dignity in Johannesburg: shortcomings of South Africa's urban refugee policy. Journal of refugee studies, 19(3), pp.308-327.
Lange, F. and Topel, R., 2006. The social value of education and human capital. Handbook of the Economics of Education, 1, pp.459-509.
Liebig, T. and Spielvogel, G., 2021. Residential segregation of immigrants: Patterns, drivers, effects and policy responses.
Martén, L., Hainmueller, J. and Hangartner, D., 2019. Ethnic networks can foster the economic integration of refugees. Proceedings of the National Academy of Sciences, 116(33), pp.16280-16285.
McPherson, M., Smith-Lovin, L. and Cook, J.M., 2001. Birds of a feather: Homophily in social networks. Annual review of sociology, pp.415-444.
McQuaid, S. L., & Singh, S. N. (2015). Ethnic residential segregation, employment and economic outcomes: A study of five minority groups in England and Wales. Population, Space and Place, 21(4), 394-411
Moscovici, S., 1973. Foreword Health and Illness: A social psychological analysis. Ed. by C. Herzlich. L. Academic Press. P. ix-xiv. Moscow State University, Series, 14, pp.19-28.
Mosselson, J., Morshed, M.M. and Changamire, N., 2017. Education and wellbeing for refugee youth. Peace Review, 29(1), pp.15-23.
Mouw, T. (2003). Social capital and fnding a job: Do contacts matter? American Sociological Review, 68(6), 868. https://doi. org/10.2307/1519749
Netto, G., 2011. Strangers in the city: Addressing challenges to the protection, housing and settlement of refugees. International Journal of Housing Policy, 11(3), pp.285-303.
Newbold, K.B. and McKeary, M., 2018. Journey to health:(Re) Contextualizing the health of Canada’s refugee population. Journal of Refugee Studies, 31(4), pp.687-704.
Oltmer, J., 2016. Germany and global refugees: A history of the Present. CESifo DICE Report, 14(4), pp.26-31.
Palloni, A., Massey, D.S., Ceballos, M., Espinosa, K. and Spittel, M., 2001. Social capital and international migration: A test using information on family networks. American journal of sociology, 106(5), pp.1262-1298.
Phillips, D., 2006. Moving towards integration: the housing of asylum seekers and refugees in Britain. Housing Studies, 21(4), pp.539-553.
Phillimore, J., 2021. Refugee-integration-opportunity structures: shifting the focus from refugees to context. Journal of Refugee Studies, 34(2), pp.1946-1966.
Platts‐Fowler, D. and Robinson, D., 2015. A place for integration: refugee experiences in two English cities. Population, Space and Place, 21(5), pp.476-491.
Saideman, S.M., 2002. The Power of the Small. SAIS Review (1989-2003), 22(2), pp.93-105.
Schüller, S.C., 2022. Ethnic enclaves and immigrant economic integration.
Tajfel, H., & Turner, J. C. (1979). An integrative theory of intergroup conflict. In W. G. Austin, & S. Worchel (Eds.), The social psychology of intergroup relations, pp. 33–47.
Tanis, K., 2018. Regional distribution and location choices of immigrants in Germany. Regional Studies.
Tanis, K. (2020). Regional distribution and location choices of immigrants in Germany. Regional Studies, 54(4), 483–494. https://doi.org/10.1080/00343404.2018.1490015
Tomlinson, F., 2010. Marking difference and negotiating belonging: Refugee women, volunteering and employment. Gender, Work & Organization, 17(3), pp.278-296.
Tumen, S., 2016. The economic impact of Syrian refugees on host countries: Quasi-experimental evidence from Turkey. American Economic Review, 106(5), pp.456-60.
Türk, V., 1992. Das Flüchtlingshochkommissariat der Vereinten Nationen (UNHCR). Duncker & Humblot.
Van Tubergen, F. and Kalmijn, M., 2005. Destination-language proficiency in cross-national perspective: A study of immigrant groups in nine western countries. American Journal of Sociology, 110(5), pp.1412-1457.
Ward, C., Bochner, S. and Furnham, A., 2020. The psychology of culture shock. Routledge.
Wagner, W., Duveen, G., Farr, R., Jovchelovitch, S., Lorenzi‐Cioldi, F., Marková, I. and Rose, D., 1999. Theory and method of social representations. Asian journal of social psychology, 2(1), pp.95-125.

Download PDF

Version 1

posted

You are reading this latest preprint version

Are refugees better together? a study on the influence of ethnic clusters and dispersal policy on refugee integration

Status:

Version 1

Abstract

Figures

1.0 Introduction

2.0 Background

2.1 Germany and refugees

2.2 Refugee allocation system and the random factor

2.3 Refugee Integration Act 2016

3.0 Literature review and research contribution

3.1 Theoretical underpinning

3.2 Empirical studies

3.2.1 Refugee integration

3.2.2 Ethnic concentration

3.2.3 Other factors

3.3 Research contribution

3.4 Research questions and objectives

4.0 Methodology and methods

4.1 Data

4.2 Identification strategy

4.2.1 Difference in difference strategy

4.2.2 Interrupted time series (ITS) analysis

5.0 Results and discussion

5.1 The effect of refugee influx and ethnic clusters on integration

5.1.1 Descriptive statistics

5.1.2 Benchmark analysis

5.1.4 Robustness checks

5.1.4.1 Propensity score matching difference in differences (DID)

5.1.4.2 Falsification test

5.1.4.3 Sensitivity test

5.1.4.4 Placebo test

5.2 The effect of the dispersal policy of the Integration Act 2015

5.2.1 Descriptive statistics

5.2.2 Benchmark analysis

5.2.3 Robustness checks

6.0 Implications and conclusion

Declarations

References

Status:

Version 1