Past and Future Trends in Refugee Migration on the Regional Level in Germany – An Analysis and Projection of Labor Market Effects Past and Future Trends in Refugee Migration on the Regional Level in Germany – An Analysis and Projection of Labor Market Effects* Patrizio Vanella, Timon Hellwagner, Philipp Deschermeier Abstract: Since 2013, more than two million refugees have arrived in Germany and have been allocated across federal states and districts according to legal policies. A steadily increasing number of refugees is now entering the German labor market, albeit under varying economic and demographic contexts. However, regional differences in refugees’ labor market integration have received little attention both retrospectively and particularly prospectively, given the projected population decline across Germany. Addressing this apparent shortcoming in the literature, we collect data on refugee arrivals by gender, nationality, approval rates, and regional allocation from 1995 to 2019. Applying principal component analysis and time series analysis, we fi rst analyze past patterns of refugee migration to Germany and project both arrivals and regional allocations by gender and nationality until 2030. Then, combining the collected migration fi gures for German labor market regions and offi cial labor market statistics, we investigate past regional employment effects from 2008 to 2019. Next, we calculate corresponding future employment effects conditional on our projected refugee fi gures, our estimation results, and offi cial regional demographic forecasts until 2030. Our fi ndings suggest that refugee migration does not affect German labor market regions equally, but instead has and will continue to lead to distinct regional employment effects. Moreover, the labor market integration differs by gender and origin of the refugees. Consequently, the interaction of regional employment effects with projected population change gives rise to different regional mitigation potentials in view of the upcoming population decline. Keywords: Refugee Integration · Regional labor markets · Population decline · Principal component analysis · Time series analysis Comparative Population Studies Vol. 47 (2022): 443-484 (Date of release: 07.12.2022) Federal Institute for Population Research 2022 URL: www.comparativepopulationstudies.de DOI: https://doi.org/10.12765/CPoS-2022-17 URN: urn:nbn:de:bib-cpos-2022-17en4 * This article belongs to a special issue on "Refugee Migration to Europe – Challenges and Potentials for Cities and Regions". ** This article has a Data Appendix with supplementary material URL: http://www. comparativepopulationstudies.de/index.php/CPoS/article/view/440/365. http://www.comparativepopulationstudies.de https://doi.org/10.12765/CPoS-2022-17 http://www http://www.comparativepopulationstudies.de/index.php/CPoS/article/view/440/365 • Patrizio Vanella, Timon Hellwagner, Philipp Deschermeier444 1 Introduction Germany, like numerous other countries, has seen a surge in refugee migration over the past years. The causes of why refugees are forced to leave their home countries are complex and are often connected to impending or implemented restrictions of individual freedoms and future insecurity (EASO 2016). The latter is also a growing consequence of nutritional problems caused, among others, by global warming (UNHCR 2020). In host countries, these recent developments have been accompanied by wide-ranging debates on refugee integration and the possible consequences for host communities. For example, some advocates of more restrictive migration policies argue that higher migration numbers might increase crime or terrorism threats (see, for instance, Vanella/Deschermeier 2018; Galantino 2022 on those discussions) or put an additional strain on social security systems (Eckert 2017). Similarly, the labor market effects of refugee migration are subject to discussions both within society and academia. In the scientifi c community, a large body of the literature has analyzed the multitude of reasons refugees enter the labor market more slowly compared to natives and other migrants and are subject to worse outcomes, including lower wages (Brücker/Jaschke/Kosyakova 2019). Explanations include worse access to and quality of education in their destination country (Hunkler et al. 2021), the complicated systems for recognizing and confi rming formal qualifi cations (BMBF 2020), and language skills (De La Rica et al. 2015). Moreover, the demographic characteristics of refugees are important for their potential labor market integration. Females have a lower ceteris paribus (c.p.) labor market participation than males. Furthermore, the refugees’ age may also have a substantial impact on their labor market integration (Gustafsson et al. 2017), as labor force participation rates are highly age-dependent (Fuchs et al. 2018). Meanwhile, refugees prevalently arrive in host countries such as Germany, where demographic developments marked by low fertility as the major driver of natural population decline (Vanella/Deschermeier 2020) are expected to cause long-term decreases in the labor force (Fuchs et al. 2018; Vanella/Deschermeier 2019). Furthermore, recent projections indicate that this overall population decline will not take effect at the same magnitude across German regions (Maretzke et al. 2021). This raises the question to what extent refugees’ labor market integration might contribute to mitigating regional labor supply decline. However, this question has not yet been addressed in the literature. While most research addresses the labor market integration–refugee migration nexus at the individual level, we analyze the potential regional variation in employment effects of refugee migration across German labor market regions in the context of population decline from a macro perspective. Detailed and longitudinal data on refugee migration are diffi cult to obtain for Germany. Therefore, as a fi rst step, we construct time series of national refugee immigration data and derive estimates for the subnational level from these. We will explain the approach in the next section. Second, we combine constructed time series with offi cial labor market statistics in an econometric model to derive Past and Future Trends in Refugee Migration on the Regional Level in Germany • 445 long-term trends of future refugee migration and quantify its past regional labor market effects, conditional on sex and nationality. These models are then combined to project the surplus number of employees eligible for social security as a result of observed and expected refugee immigration through 2030. Put differently, we combine an econometric labor market model and a demographic forecast model to project the employment effect of refugee migration for 34 labor market regions in Germany by 2030. The remainder of the paper is structured as follows. Following a short illustrative literature review of refugee migration projections and the connection between refugee migration and labor market integration (Section 2), we outline the baseline data of our study and our methodological approach (Section 3). Next, we present the study’s results (Section 4). After a discussion of our fi ndings and the paper’s limitations (Section 5), we outline the need for further, more detailed data and delineate avenues of future research (Section 6). 2 Literature review 2.1 Refugee migration projections For destination countries such as Germany, refugee migration constitutes an important share of overall migration (Vanella/Deschermeier 2018), which, in turn, is an important component of long-term population forecasts (Vanella/Deschermeier 2020) that are indispensable for economic and infrastructural planning (Vanella et al. 2020). Nevertheless, the literature on the estimation of future refugee migration is rather scarce. To our knowledge, no long-term refugee migration projections have been conducted to date, with the exception of refugee migration associated with climate change (UNHCR 2020). For instance, UNHCR (2019) only projects forced migration by target country two years out. Longer forecast horizons are discussed qualitatively, but no quantitative results are presented. For example, UNHCR (2020) expects a c.p. increase in globally displaced persons by 2050 due to global warming and related increases in extreme weather events, especially in Sub-Saharan Africa. Most forecasts of international migration do not address refugee migration at all (see, e.g., Sohst et al. 2020 for a recent systematic review of migration projections). Those stressing the importance of refugee migration in forecasting future migration fl ows only address the issue indirectly in their forecasts (see, e.g., for Germany Deschermeier 2016 or Fuchs et al. 2018). For instance, Bijak et al. (2019) discuss a separate Bayesian time series approach for forecasting infl ows of asylum seekers in the United Kingdom (UK). However, the authors do not fi nd the results satisfactory. Vanella and Deschermeier (2018) indirectly include refugee migration by demographics to Germany in their forecast model of net migration via principal component analysis (PCA). They interpret one of the major principal components (PCs) as a crises index, which covers the impact of crises on (refugee) migration. However, their model only covers asylum migration indirectly as a residual which cannot be interpreted directly. Given that refugee migration is subject to • Patrizio Vanella, Timon Hellwagner, Philipp Deschermeier446 unforeseeable events such as sudden violent confl icts,1 it is hardly surprising that no long-term projections have been conducted, since such shocks are very diffi cult to predict in the long term (Deschermeier 2016). 2.2 Refugee migration and labor market outcomes A vast number of studies analyze refugees’ labor market outcomes, representing a steadily increasing part of the economic literature on the migration-labor market nexus.2 First, the empirical evidence3 suggests that, for developed countries, the overall employment share among migrants is lower compared to natives, with converging shares over time. The difference increases further for refugees (refugee gap) (e.g., Brell et al. 2020; Dustmann et al. 2017; Fasani et al. 2018; Brücker/ Jaschke/Kosyakova 2019). Second, studies have found occupational differences between migrants and natives. Dustmann/Frattini (2013) show that immigrants are more likely to work in occupations requiring a lower level of skills. This occupational migrant-native gap is confi rmed by Fasani et al. (2018) who demonstrate that the differences are more distinct for refugees. Notably, there is substantial occupational underemployment among refugees who arrived in Germany between 2014 and 2016 (Brücker/Jaschke/Kosyakova 2019). Third, fi ndings suggest that migrants’ wages are lower. Dustmann/Frattini (2013) provide evidence that across European countries, migrants are more likely to be in the bottom decile of the earnings distribution than natives. Once again, there is evidence of a more distinct refugee gap (e.g. Brell et al. 2020). Various approaches seek to explain such differences in labor market outcomes, with education generally being considered as a key factor. Among the refugee population in Germany, heterogeneous educational patterns emerge: First, as Brücker/Jaschke/Kosyakova (2019) note, the human capital levels among refugees in Germany exhibit large shares at both ends of the educational continuum, that is, either a secondary education or above, or little to no formal education at all. Second, as Guichard (2020) shows, the educational background and self-selection of refugees concerning education varies by country of origin. Basilio et al. (2017) provide corresponding evidence for Germany. Here, missing formal recognition of education obtained abroad may explain a substantial part of the differences (Brücker et al. 2021). Another factor contributing to the integration and outcomes of migrants is language skills (De La Rica et al. 2015). Indeed, empirical evidence suggests that 1 We will discuss this in more detail for the case of Germany in Section 3.1. 2 A comprehensive overview concerning migrants’ labor market integration can be found in De La Rica et al. (2015). With a narrow focus on the labor market outcomes of refugees, the recent survey by Becker/Ferrara (2019) offers a broad view. 3 If not indicated differently, the quoted studies and respective results rely on conditional analyses incorporating individual characteristics, such as education, to account for aggregate differences in native, migrant, and refugee populations. We outline the most important of these characteristics below. Past and Future Trends in Refugee Migration on the Regional Level in Germany • 447 lacking profi ciency in the host country’s language might explain a large part of the (refugee) employment gap (Brell et al. 2020). Furthermore, (mental) health is comparatively worse among refugees than other migrants (Brell et al. 2020) due to traumatic experiences during the journey (see, e.g., Schock et al. 2016). In labor market research, addressing the role of health for labor force participation has already found a widespread application (e.g., Cai 2010). The results presented by Ruiz/Vargas-Silva (2018) indicate that this might explain parts of the refugee gap. 2.3 Towards a regional perspective Labor market outcomes are not only determined by (observable4) individual characteristics but also by the broader social and economic contexts refugees are embedded in. For example, the institutional setting infl uences labor market integration and outcomes. In Germany, labor market admission is regulated by federal law.5 Importantly, the waiting time for employment opportunities has been reduced from twelve to three months in 2014 but, contrarily, asylum seekers from “safe countries of origin” no longer have access to employment (Brücker/Jaschke/ Kosyakova 2019). Empirical evidence indicates an adverse effect of employment bans (e.g., Marbach et al. 2018). Moreover, the literature suggests that labor force participation rates move cyclically – increasing during times of economic growth, declining during recessions (e.g. Aaronson et al. 2014). This is particularly interesting since empirical evidence suggests that the business cycle disproportionally infl uences the labor market outcomes of migrants. For example, Dustmann et al. (2010) show that economic shocks impact the unemployment rates among immigrants in Germany more adversely compared to natives within the same skill group. As other high-income countries, Germany is expected to experience unprecedented demographic developments in the near future: Aside from further population aging (Vanella/Deschermeier 2020), its (potential) labor force will decline substantially (Fuchs et al. 2018). As in the past, positive net migration and, more importantly, rising labor force participation rates and digitization could mitigate these developments, but hardly stop them (see, e.g., Brenke/Clemens 2017 or Fuchs et al. 2019). Regarding age and gender structure, the migrant and refugee populations differ from the total foreign and overall population in Germany to varying degrees (see Destatis 2020a). Even though the refugee population is substantially younger, this may not proportionally increase the labor force. While participation rates of males between 15 and 65 years are similar for natives and foreigners (83.9 percent vs. 81 percent), this is not the case for females (77.1 percent vs. 60.8 percent) (Destatis 2020b). 4 In the literature, other explanatory factors which are less easily observable (such as discrimination) are discussed as well. As stated in the beginning, we do not provide an exhaustive literature overview of the comprehensive research on the migration-labor market nexus but refer to corresponding papers noted above. 5 §61 Asylum Act (AsylG). • Patrizio Vanella, Timon Hellwagner, Philipp Deschermeier448 However, the outlined dimensions do not necessarily affect refugees’ labor market integration uniformly across Germany. First, recent projections show very different demographic developments at the regional level until 2040, with some regions (districts) expected to face further population growth while others are likely to experience a substantial decline, in particular among those of working age (see Maretzke et al. 2021). Second, there is evidence that labor force participation rates differ regionally (see, e.g., Wanger 2020). Third, labor market forecasts indicate that the regional labor demand is expected to vary both by economic sector and skill level in the upcoming decades (Zika et al. 2020). Fourth, in Germany, labor market admission is regulated by federal law, thus there are no formal differences across the country. However, the formulation of the federal law gives rise to a three-stage labor market admission procedure, which alters the regional institutional setting for refugee integration as described by Brücker/Jaschke/Kosyakova (2019). Additionally, dispersal policies coupled with residency restrictions might play a crucial role. In Germany, arriving refugees are distributed across the 16 federal states according to an agreement called Königsteiner Schlüssel (KS). In the literature, the coupling of dispersal policies and residency restrictions are found to impose negative effects on the labor market outcomes of refugees (e.g., Fasani et al. 2018 or Brücker et al. 2020). Regional variation of all the discussed aspects – from demographic changes to the institutional setting, from structural differences of regional labor markets to refugees’ individual characteristics and internal migration – give rise to the possibility of regionally varying employment effects, in particular in the context of an aging and decreasing population, that is labor supply decline. Yet, to the best of our knowledge, this issue has not been addressed in the literature. Thus, as a working hypothesis, based on the literature review and the argumentation above, we propose that the regional interaction of employment dynamics and demographic changes yield regionally varying mitigation potentials for labor supply decline. As noted in the introduction, while the literature outlined above is strongly focused on micro-level studies, i.e., analyzing the labor market integration of refugees from an individual perspective, we complement the existing research by approaching employment effects from a regional, macro-level point of view. 3 Data and methods 3.1 Forecasting refugee migration and asylum requests 3.1.1 Refugee migration to Germany since 1995 – trends and background Our analysis follows a hierarchical approach. As the fi rst step of the analysis, we investigate long-term trends of refugee migration at the national level. For this, we use annual data on initial asylum requests (IARs) in Germany by group of origin, provided by the German Federal Offi ce for Migration and Refugees (Bundesamt für Migration und Flüchtlinge – BAMF) and the Federal Ministry of the Interior, Building Past and Future Trends in Refugee Migration on the Regional Level in Germany • 449 and Community (Bundesministerium des Innern, für Bau und Heimat – BMI) in their annual migration reports (BAMF 2020b; efms 2003). Because the defi nition of IARs was introduced in 1995 (BAMF 2017), we choose the period 1995-2019 for our analysis. To retain enough statistical power, we aggregate the exact nationalities of asylum seekers into four origin groups: European, African, Asian, and others. We analyze the long-term trends in IARs in Germany for these four defi ned origin groups. Figure 1 shows the time series of the four origin groups for the baseline period. There appear to be high correlations in application numbers among the four origin groups. However, trends in applications by persons with European citizenship tend to precede the other nationality groups by one year. After generally decreasing trends until the mid-2000s, asylum requests started increasing in 2007. Table A1 gives an overview of major refugee infl ows to Germany since then, with information on the major origin countries and causes for emigration. Among all nationalities, the number of IARs from 2011 to 2012 more than doubled from 11.000 to 23.000. In sum, annual increases in new asylum requests in Germany are observable between 2007, the starting year of the fi nancial crisis, and 2016, from 19.000 to 722.000 IARs. In particular, the sharp increases in the mid-2010s are caused by the growing reach and violence of various terror groups, most notably the Islamic State in Syria and Iraq, the Taliban in Afghanistan, and Boko Haram in West Africa (Heidelberg Institute for International Confl ict Research 2016). Since 2017, the number of requests has declined after the EU initiated a series of measures Fig. 1: Annual initial asylum requests by origin group in thousands 0 100 200 300 400 500 600 1995 2000 2005 2010 2015 Year European African Asian Others 0 100 200 300 400 500 1995 2000 2005 20152010 Initial asylum requests (in k) Source: efms 2003; BAMF 2020b; authors’ calculation and design. • Patrizio Vanella, Timon Hellwagner, Philipp Deschermeier450 which made it more diffi cult to enter the EU at its eastern borders (SVR Integration und Migration 2018). 3.1.2 Forecast model of initial asylum requests Since our goal is not only to describe past trends in asylum applications in Germany, but to also predict their future development, we must consider the observed correlations between the different time series in our analysis. This is done via principal component analysis (PCA), which transforms the original, correlated variables into the same number of uncorrelated, new variables, which are linear combinations of the original variables. PCA chooses the principal components (PCs) such that most high dimensional problems can be covered well by a small number of index variables. PCA moreover allows for including the correlations between the original variables in simulation studies.6 The method has been applied to forecasting international migration in Germany by Fuchs et al. (2018) and Vanella and Deschermeier (2018). The fi rst PC of the asylum request time series ( 1p A) explains 86.5 percent of the variance in the past observations of the four origin-specifi c asylum request series. Its loadings are negative for all four series (-0.4 for European, -0.45 for Africans, -0.51 for Asians, -0.62 for others); therefore, increases in 1p A are associated with c.p. decreases in the number of IARs in Germany from all nationality groups. Therefore, we interpret 1p A as a general Refugee Migration Index. The hypothetical course of the Refugee Migration Index with its projected future course is illustrated in Figure 2. The Refugee Migration Index mirrors the overall trends in asylum requests, with increases (i.e., c.p. decreases in asylum requests) until 2008 and rather sharply decreases (i.e., c.p. increases in asylum requests) until 2016, after which we observe slight increases due to the EU measures aimed at containing refugee migration, as mentioned in Section 3.1.1. No clear long-term trend can be identifi ed, as the index is highly infl uenced by current crises, which are diffi cult to predict. We therefore follow a similar assumption as in Vanella and Deschermeier (2020, 2018), who interpreted a similar index as a general Crises Index in their migration forecast model, stating that in the long term 1p A is expected to converge towards its median over the observed period. However, this assumption is relaxed by fi tting an AR(1) process to the past data. Thus, the slope of the curve is computed from the past data, not based on further assumptions. The resulting model for year y is (1) generates the prediction of 1p A until 2030. For the other four PCs, we assume random walk processes without drift in the future, which implicitly predicts the value of, say 𝔼[ 2p A (y + 1)] to equal the 6 For a more detailed description of PCA and its application in demographic forecasting, see Vanella (2018). | 1 18.192 0.885 1 (1) Past and Future Trends in Refugee Migration on the Regional Level in Germany • 451 observation 2p A (y). The random walk’s expectation is therefore similar to the naive prediction. 3.2 Refugee demographics For Germany, combined data on demographics and origin of refugees are not available. However, we can gather data on age- and sex-specifi c information for the sum of the IARs, without disaggregation by nationalities. For this, we use BAMF reporting data on the demographic structure of persons requesting asylum annually from 2009-2019 (BAMF 2010, 2011, 2012, 2013, 2014, 2015, 2016, 2017, 2018, 2019a, 2020a). Appendix Table A2 displays the shares of asylum seekers by age-sex group for this period. We tested these data on the national level for trending behavior in asylum requests. However, a detailed differentiation by age group led to very small numbers of refugees by stratum on the NUTS-3 level (see Section 3.5 for details). An investigation of the trends in sex-specifi c requests, however, implies a connection of the share of males to the total level of IARs. As the time series is relatively short, deriving long-term trends via time series methods for the demographic groups would yield misleading estimates.7 Following our assumption Fig. 2: Historic development and prediction of the Refugee Migration Index -23 -22 -21 -20 -19 -18 -17 -16 -15 1995 2000 2005 2010 2015 2020 2025 2030 Year -22 Refugee migration index -21 -20 -19 -18 -17 -16 Source: Authors’ computation and design. 7 We tried this in an earlier approach. • Patrizio Vanella, Timon Hellwagner, Philipp Deschermeier452 that the IARs eventually converge to their long-term median, a similar assumption appears plausible for the demographic distribution among the refugees if these phenomena are connected. Therefore, our prediction follows an AR(1) model for both variables, similar to (1). The model for the share of males below age 65 among all IARs is the corresponding model for females is Note that this model does not explicitly include correlation among the shares of both genders. In our deterministic approach, this is acceptable. More sophisticated approaches should account for this, however. (2) and (3) are used as backcast functions to approximate the missing sex-specifi c data before 2009 and as a forecast function for the years 2020-2030. 3.3 Estimated distribution of asylum seekers among federal states Refugees are distributed among the German federal states following the KS (see 2.3), which takes the population size and the economic strength of the federal states into account. Annual values of the KS from 1995-2018 were provided by the German Joint Science Conference (Gemeinsame Wissenschaftskonferenz – GKW), partly online (GKW 2021a), partly upon request (GWK 2021b). To cover correlations between the time series of the KS, we apply PCA to the logarithmized KS values. Since the annual sum of all KS values is one, we need to include one condition. As Bremen is the federal state with the smallest population, we model its refugee distribution indirectly by subtracting the sum of all other shares from one: The fi rst PC explains 93.9 percent of the variance in the refugee distribution among the federal states, therefore we call it the KS Index. The loadings of the KS Index for the 15 included federal states ordered by the offi cial geographic numbering are illustrated in Figure 3. The accompanying information on the federal states can be found in Figure 8. The federal states with positive loadings represent eastern Germany, the former German Democratic Republic (GDR) (states 12-16), except Berlin (11). Interestingly, Saarland at the French border and the most populous federal state, North Rhine- Westphalia in the West of Germany, are positively correlated to the KS Index also. | 1 0.629 0.734 1 , (2) | 1 0.362 0.753 1 . (3) 1 (4) Past and Future Trends in Refugee Migration on the Regional Level in Germany • 453 For a deeper interpretation, we must consider its past trend, which is illustrated in Figure 4, with its predicted future development as explained below. The KS Index shows a clear downward trend over the baseline period, meaning that there are clear trends in the distribution of refugees among the federal states. We extrapolate the trend until 2030 by fi tting the following model: with 1u K (y − 1) being the residual of the forecast from the preceding year 1u K (y − 1) ≔ 1p K (y − 1) 𝔼[ 1p K (y − 1)| 1p K (y − 2)]. Figures 5 and 6 illustrate the annual population growth of the 16 federal states from 1995-2018. Ignoring the countrywide decrease in 2011, when the German population count was corrected downwards by 1.3 million in the census-based estimate (Vanella et al. 2020), we see a connection between the KS Index and population growth. The federal states in Figure 5 have a negative correlation with the KS Index and witnessed positive trends in population growth since the mid-1990s, whereas the opposite is the case for the federal states in Figure 6. The exception is North Rhine-Westphalia (5), whose long-term population growth does not follow a clear trend but has a positive tendency. However, there is a clear negative long- term trend in North Rhine-Westphalia’s population share. An explanation could be the state’s comparatively low economic growth. As economic development is Fig. 3: Loadings of the fi rst principal component of the KS model -0,4 -0,2 0,0 0,2 0,4 0,6 0,8 1 2 3 5 6 7 8 9 10 11 12 13 14 15 16 Federal State -0.2 Loading 0.0 0.2 0.4 0.6 Source: Authors’ calculations and design. | 1 27.297 0.016 1 , (5) • Patrizio Vanella, Timon Hellwagner, Philipp Deschermeier454 connected to tax income originating from the federal states (Schulte 2015), differing economic developments are also mirrored in the KS Index. The remaining 14 PCs are assumed to follow random walk processes. The results of the projection of the refugee distribution among the federal states are presented in Section 4. 3.4 Trends in asylum decisions by groups of origin The share of asylum seekers who received a protection status by group of origin is derived from the offi cial data on asylum decisions from 1998-2020 provided to us by the BAMF (2021) upon request. There is a variety of different classifi cations of residence permits in Germany, which defi ne whether a person is allowed to remain in the country permanently, for a limited time, or must leave the country; this defi nes the amount of fi nancial aid refugees receive.8 As a thorough analysis of this would exceed the scope of our paper, we refer to the simple binary classifi cation whether asylum seekers are allowed to stay in the country for a (limited) period or must leave immediately. The share of persons allowed to remain in Germany is summarized by the protection rate. Figure 7 illustrates the historical development of the protection rates for the defi ned four groups of origin. Fig. 4: Past course and prediction of the KS Index -5.3 -5.2 -5.1 -5.0 -4.9 -4.8 -4.7 1995 2000 2005 2010 2015 2020 2025 2030 Year Koenigsstein Index Source: Authors’ computation and design. 8 An overview, albeit in German, is given, e.g., in BAMF 2019a/b. Past and Future Trends in Refugee Migration on the Regional Level in Germany • 455 Whereas the protection rates of European citizens had a downwards trend until the mid-2010s, we observe a sharp increase since 2016, mainly driven by high and increasing protection rates among Turkish citizens (45.2 percent in 2020). This trend was caused by a failed military coup in the summer of 2016, after which many suspects fl ed the country. Moreover, the Turkish government has restricted journalism in the country ever since, leading to journalists and opposition fi gures fl eeing the country (Zeit 2019). For all other groups of origin, besides some fl uctuations, the protection rates show an increasing long-term trend since the mid-2000s. The protection rate of Asian citizens in particular peaks considerably in 2008. This development was caused by the harmonization of the EU asylum system, which was supposed to facilitate the process of refugee protection in the EU (Haase/Jugl 2007; Engler/ Schneider 2015; EU 2011). Protection rates are, although infl uenced by stochastic phenomena, primarily driven by political decision-making and are therefore by nature deterministic. Therefore, we will not construct a forecast model for them but rather assume a naive scenario which fi xes the protection rates at their 2020 values until 2030, i.e., 25.4 percent for Europeans, 27.9 percent for Africans, 60.7 percent for Asians, and 53.9 percent for all others. Fig. 5: Annual population growth of federal states with negative loadings in the KS Index -50 0 50 100 150 200 2000 2005 2010 2015 Year 1 2 3 6 7 8 9 11 0 50 100 150 Absolute population growth (in k) Census Source: GENESIS-Online 2021; authors’ computation and design. • Patrizio Vanella, Timon Hellwagner, Philipp Deschermeier456 3.5 The regional distribution of asylum seekers Refugees assigned to specifi c federal states are distributed to the NUTS-3 level (districts). In most cases, the distribution of refugees is based on the population size of the district relative to the federal state (Geis/Orth 2016). As population data in Germany are normally published with a lag of over one year, we assume the second lag of the annual population estimates as the basis for the distribution of refugees to the NUTS-3 level. For this, the end-of-year population numbers for 1995- 2019 are downloaded from the database of the federal and state statistical offi ces (Statistische Ämter des Bundes und der Länder 2021). To estimate the distribution of refugees in the districts within the federal states, we assume annual distribution according to the most recent regional population projection of the Federal Institute for Research on Building, Urban Affairs and Spatial Development (Bundesinstitut für Bau-, Stadt- und Raumforschung – BBSR). The data was provided upon request by the BBSR (2021). We use these BBSR estimates for the future distribution of Fig. 6: Annual population growth of federal states with positive loadings in KS Index, including Bremen -50 0 50 100 150 200 2000 2005 2010 2015 Year 4 5 10 12 13 14 15 16 0 50 100 150 Absolute population growth (in k) Census Source: GENESIS-Online 2021; authors’ computation and design. Past and Future Trends in Refugee Migration on the Regional Level in Germany • 457 refugees within the federal states to the district level. We interpret this distribution as politically driven and therefore deterministic.9 In our analysis, we investigate the effects of refugee migration in 34 different labor market regions (LMRs) as suggested by Zika et al. (2020), who, in turn, draw on the original classifi cation by Kropp and Schwengler (2011, 2016). The selection of LMRs for analyses in a regional perspective is widely used in the literature (e.g. Burstein et al. 2020). A functional delineation of this kind appears to be preferable to administrative borders since LMRs capture most commuting fl ows. In other words, residence in the small-area, district level is not necessarily indicative of whether individuals work in the respective region as well, a point of particular importance in the context of this analysis. However, employment within the LMR of residence can be assumed. The original delineation by Kropp and Schwengler (2016) correctly estimated about 90 percent of labor commuters in Germany to their respective LMRs. The aggregation by Zika et al. (2020), which primarily aims at creating LMRs with suffi cient case numbers for projections, further increases this share. Appendix Table A3 gives an overview of the districts included in each LMR. Figure 8 gives a detailed overview of the LMRs Fig. 7: Protection rates in Germany by group of origin 0 20 40 60 80 2000 2005 2010 2015 2020 Year European African Asian Others Protection rate (in %) Source: BAMF 2021; authors’ computation and design. 9 In fact, a refugee distribution linked to the more granular units makes it highly stochastic, as the future population size per se is highly volatile, which, in relative terms, holds especially true for relatively small regional populations (Deschermeier 2011; Vanella/Deschermeier 2020). • Patrizio Vanella, Timon Hellwagner, Philipp Deschermeier458 and their composition from the NUTS-3 to the NUTS-0 level. The expected annual distribution of refugees below age 65 by sex and group of origin among the LMRs, as derived from our projection, is provided in the Online Data Appendix. 3.6 Estimation of labor market effects and projection In the fi nal part of our analysis, we investigate the regional labor market effects of refugee migration, drawing on our previous results and using the LMR classifi cation outlined above. Given the availability and high dimensionality of regional data and for the sake of a consistent and parsimonious empirical approach throughout our analysis, we rely on PCA. We apply separate PCAs on the annual differences in regionally employed persons from 2007 to 2019 (annual average) for each gender- nationality group, yielding six10 individual models. In doing so, we account for Fig. 8: Overview of the labor market regions as used in the analysis Source: Authors’ illustration using BKG 2021. 10 We excluded refugees, both male and female, with European citizenship from this part of our analysis. The reliable empirical assessment of their labor market effect is strongly hampered by various factors such as enlargements of the EU. Increased labor-related migration from former refugee-sending countries to Germany because of this enlargement process vastly outperforms the number of asylum requests in the preceding years. Past and Future Trends in Refugee Migration on the Regional Level in Germany • 459 correlations in the regional employment by including simultaneous employment trends in the LMRs in the analysis. As it is diffi cult to measure interregional migration of persons with an asylum background, PCA indirectly quantifi es the gravity of the LMRs for asylum seekers and hence their movements between the LMRs. Moreover, the method enables the modeling of labor market trends for all LMRs by a small number of indices instead of requiring a multitude of separate models for each LMR. The employment data are taken from the statistical database of the Federal Employment Agency (FEA, Bundesagentur für Arbeit – BA) and refer exclusively to employees subject to social insurance. In each gender-nationality case, the fi rst of the resulting 34 PCs explains over 89 percent of the variance in the employment differentials. Thus, we regress each fi rst PC, pcgn,t, on the gender-nationality specifi c refugee fi gures, refgn,t, and the real gross domestic product (GDP) per capita growth rate, gdpt. More formally, our model can thus be written as pcgn,t = c + refgn,t + gdpt +et where c is an intercept and et is the error term. The refugee fi gures refer to the aggregated number of the past seven years since survey data indicates that refugees’ labor market integration levels out seven years after arrival (Brücker/ Jaschke/Kosyakova 2019). The real GDP per capita was included as a predictor of overall economic conditions, which, as outlined in Section 2, plays a key role in the labor market participation and economic success of foreigners in general. GDP data was taken from VGRdL (2019). Using the obtained coeffi cients from (6), we derive both in-sample and out-of- sample projections for each fi rst PC (see Table A5). For in-sample projections, we draw on observed refugee migration and GDP values. For out-of-sample projections, we assume GDP growth according to the compound annual growth rate from 2010 to 201911 and rely on the projected refugee fi gures as outlined above. In both cases, the remaining 33 PCs are again assumed to follow random walks. Applying matrix algebra on the newly derived matrix of PCs and the matrix of eigenvectors, we derive annual regional employment fi gures. In a second step, we repeat this exercise but exclude refugee migration in the calculation of the fi rst PC. By forming differences between both projections with and without refugee migration, we derive the annual regional employment impact for the past and the future. We cross-validate the plausibility of our empirical strategy by comparing the in- sample projection results for 2019 to FEA employment stock statistics in Figure 9. The offi cial FEA statistics refer to the stock of employees eligible for social insurance who are labeled as individuals in the context of refugee migration in March 2020, which is the fi rst available record of this statistic. This defi nition encompasses all those persons whose residence status is related to an asylum procedure. Both statistics are shown per 100,000 inhabitants (2019 fi gures from Statistische Ämter (6) 11 In doing so, we do not account for economic downturns due to the pandemic situation since March 2020. • Patrizio Vanella, Timon Hellwagner, Philipp Deschermeier460 des Bundes und der Länder 2021). As can be seen, our model replicates the regional variation in employment stock fi gures fairly well. However, LMRs such as Munich (20) or Berlin (26) show by far larger estimated absolute employment numbers compared to offi cial statistics. This is likely a result of several factors. Economically, regional structural or institutional conditions and differences thereof may not be fully captured by the estimation strategy and, thus, may explain comparatively lower or higher integration of refugees. Demographically, different regional immigration and emigration patterns may hamper a more precise inference and estimation of regional fi gures. Methodically, the model overestimates the number of people, as we are not able to account for people leaving employment, for example, due to retirement, job loss, or a higher number of job turnovers in general. Additionally, the model does not cover people who stay employed but experience changes in their residence status and thus are no longer classifi ed by offi cial FEA statistics as being “in the context of refugee migration.” Below, we account for the outlined shortcomings in terms of overestimation by correcting the estimated values for the period from 2020 to 2030, using the ratio of offi cial and estimated values shown in Figure 9. Thereby, we are implicitly assuming that the difference between annually estimated employment infl ow fi gures and true employment infl ow fi gures is stable over time. 4 Results Maintaining the outlined order of partial models from our empirical strategy, we present the corresponding results below. First, projected IAR fi gures by gender and nationality group as well as the estimated distribution across labor market regions are shown. Second, we analyze regional labor market impact: We relate the cumulated Fig. 9: Comparison of FEA stock statistics and model results 100 200 300 400 500 600 700 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 Estimation results FEA statistics Employment of Refugees: Comparison of FEA Stock Statistics and Estimation Results in 2019 Labor Market Region Em pl oy ed /1 00 ,0 00 Source: Statistische Ämter des Bundes und der Länder 2021; Federal Employment Agency 2021; authors’ calculation and design Past and Future Trends in Refugee Migration on the Regional Level in Germany • 461 estimated in-sample employment numbers in 2019 to the overall refugee infl ow fi gures in the regions, yielding an interpretation of different degrees of regional employment effects. We then compare employment and unemployment patterns, providing insights on possible regional differences in employment compared to overall labor market integration. Subsequently, we relate the projected employment fi gures until 2030 to the regional working-age population fi gures projected by the BBSR (Maretzke et al. 2021), thus offering an estimate of the contribution of refugee migration to mitigating the projected labor supply decline. First, we expect IARs to further decrease beyond 2030, as illustrated by the projected increase in the Refugee Migration Index in Figure 2. Note that due to the negative correlation between that index and the IARs shown in Figure 1, increases in the index are associated with c.p. decreases in the asylum requests. The decline is particularly distinct for Asian citizens, driven by the very high number of IARs in 2015 followed by a subsequent long-term decrease to the median level. Multiplying the estimated future IAR fi gures with the projections of the shares of males and females below age 65, we derive projections of origin- and gender-specifi c IARs. These numbers are then multiplied by the federal state shares as outlined in Section 3.3. The predicted annual distribution of the refugees to the federal states is listed in Appendix Table A4. As noted, using data of past refugee migration to Germany and drawing on our estimates of future developments outlined in the preceding paragraphs, we then investigate corresponding regional employment effects. Figure 10 shows the cumulative employed refugees according to the in-sample projection relative to the cumulative regional refugee infl ow numbers. Thus, employment fi gures refer to 2008-2019, corrected as outlined in Section 3.6, and refugee fi gures refer to 2001- 2018 as we use the cumulated fi gures of the past seven years in the model. Dark purple values indicate more people being employed relative to infl ow numbers, that is, 0.25 or more employees eligible for social security per person who had initially arrived in the respective region. Those regions illustrated by the lightest color have seen 0.15 or less employment increase per person arrived. While Frankfurt am Main (10) experienced the strongest increase (0.35), Greifswald-Stralsund (29) has seen the smallest employment effect (0.08). Notably, these results should not be interpreted as better or worse regional employment integration per se relative to immigration fi gures, but rather as regional employment receptivity to refugee migration, as the results displayed may be infl uenced by processes such as internal migration. The conceptual differences and limitations thereof are further discussed in the following section. Figure 11 in turn displays the cumulative corrected estimation results in 2019 relative to the average annual unemployment stock fi gures of persons in the context of refugee migration according to offi cial FEA statistics. Dark brown values indicate comparatively more refugees being employed than unemployed (1.25 or more). Dark turquoise areas refer to the opposite (0.75 or less). Compared to the previous map, we can observe not only a distinct East–West pattern but also a North–South divide. Munich (20) exhibits the largest value with 2.66. This means that for every • Patrizio Vanella, Timon Hellwagner, Philipp Deschermeier462 unemployed person in the context of refugee migration, 2.66 persons are employed. The lowest value is 0.52 for Greifswald-Stralsund (29). Given our interest in the contribution of employment effects by refugee migration for mitigating population and labor supply decline, the projected fi gures may be related to existing population forecasts. Here, we draw on the demographic projection by Maretzke et al. (2021), which indicates a decline of the working-age population (aged 16-64) across all labor market regions through 2030. Figure 12 illustrates the cumulative projected employment effects for 2020-2030 per expected working-age population decline in the same period, measured in absolute numbers. Thus, dark green values indicate stronger mitigation of population decline by refugee migration (more than 0.2, i.e., 20 percent), light green values refer to the opposite (less than 0.1, i.e., 10 percent). Munich (20) again exhibits the highest value among all regions with 0.27, closely followed by Berlin (26) with 0.26. Conversely, Neubrandenburg (28) shows the lowest estimated value, at 0.02. Fig. 10: Regional employment effects relative to immigration fi gures until 2019 Source: BKG 2021; Federal Employment Agency 2021; authors’ computation and design. Past and Future Trends in Refugee Migration on the Regional Level in Germany • 463 5 Discussion The results presented in the preceding section demonstrate distinct regional employment patterns regarding refugee migration, in particular in the context of population decline. However, the empirical approach applied in this paper and the corresponding results still require a detailed discussion, including their limitations. The comparisons of cumulated employment fi gures to both cumulated infl ow numbers and stock unemployment in FEA statistics, reveal stark differences across Germany. The results for regional receptivity (Fig. 10) display the extent to which a labor market region has experienced employment growth due to refugee migration. Interestingly, the differences largely run along the borders of Western and Eastern German labor market regions. The former have seen disproportionally stronger employment increases, whereas employment growth due to refugee migration has been lower for the latter. However, this observation does not necessarily imply a stronger labor market integration per se, but may be the result of only implicitly modeled processes such as internal migration after arrival. To this end, the comparison of employment receptivity to employment integration patterns (Fig. 11), the latter by relating the employment effects to stock unemployment fi gures, provides additional insights. Here, by contrast, clear differences between Fig. 11: Regional employment effects relative to unemployment stock fi gures Source: BKG 2021; Federal Employment Agency 2021; authors’ computation and design. • Patrizio Vanella, Timon Hellwagner, Philipp Deschermeier464 Western and Eastern labor market regions largely vanish, whereas a distinct North– South gradient emerges. Thus, labor market regions in the South of Germany have seen, at least in relative terms, more refugees and asylum seekers transitioning into employment. The conjunction of both analyses may point towards a twofold explanation: First, we do not model migration between regions after the arrival explicitly. As outlined in the literature review, the empirical evidence suggests that both free choice of residence and work as well as proximity to ethnic communities intervene in the integration process. Our model does not disentangle these factors and their contribution to the derived results. Detailed analyses of the distribution of refugees with approved protection status support the internal migration assumption (see, e.g., Rösch et al. 2020). Similarly, the existence of ethnic networks fosters integration. Given the sharp differences in regional foreign population shares across German regions (see, e.g., BAMF 2019c), these networks possibly emerge as another intervening mechanism. Regionally varying labor force participation rates may eventually lead to differing employment effects; however, one may assume these differences to be comparatively small. Second, the observed patterns in employment-to-unemployment ratios largely coincide with regional economic strength (see VGRdL 2019). As noted earlier, the literature suggests that migrants’ Fig. 12: Regional employment effects relative to population decline until 2030 Source: Maretzke et al. 2021; BKG 2021; authors’ computation and design. Past and Future Trends in Refugee Migration on the Regional Level in Germany • 465 employment rates are linked to economic conditions. However, given the modeling strategy, we are not able to derive defi nitive conclusions for these two explanatory approaches. Interpreting the relation of employment effects and future demographic changes, that is, the contribution of refugee migration to mitigating labor supply decline, is fairly straightforward. Relating the conditional projection of the observed past employment effects until 2030 to projected working-age population declines indicates that future employment effects strongly interact with expected demographic developments. Thus, concerning the working hypothesis in Section 2.3, we fi nd that refugee migration can indeed be expected to mitigate labor supply decline differently across regions. This is particularly visible in the examples of labor market regions surrounding large cities, most of all for Munich and Berlin, which do not belong to the group of LMRs witnessing the largest employment effects relative to immigration but can expect refugee migration to compensate for the decline in the labor supply12 by more than a quarter. Importantly, according to Maretzke et al. (2021), both regions are expected to face signifi cantly smaller declines in the working-age population than the remaining regions. For a multitude of other regions which are expected to face a more distinct decline, the mitigation potential is mostly – but not necessarily – much smaller. For example, economically strong regions in the South-West of Germany, such as Frankfurt am Main, Freiburg, and Ravensburg, with strong employment integration, i.e., large employment-to- unemployment ratios, also exhibit comparatively high mitigation potentials despite stronger population decline. By contrast, labor market regions in Western Germany with large employment effects relative to immigration, do not exhibit similarly large mitigation potentials due to sharper population decline. Moreover, the contribution of refugee migration to cushioning labor supply decline in labor market regions in Eastern Germany, given the expected intensities of population decline, remains limited, even for areas with strong employment integration such as Erfurt. Thus, the results presented above not only illustrate varying mitigation potentials but also point to regionally varying causes thereof. All these conclusions hold only if one relies on the implicit assumption that both receptivity and integration patterns as observed until 2019 do not change in the future. Put differently, altered internal migration patterns as well as converging or intensifying regional differences in employment-to-unemployment ratios among refugees may increase or decrease the mitigation potentials shown and discussed above. Further research which explicitly examines internal migration and regional unemployment patterns among refugees may offer signifi cant contributions. From a much more general perspective, we might also conclude that in the medium- to long-term, employment effects as a consequence of refugee migration, as observed and estimated in our model, could also provide diverging potentials for economic development among German labor market regions. Once again, we recommend that additional research should be done to support these fi ndings. 12 Only under the assumption that the decline of the working-age population equals or at least implies a proportional decline in the labor force and employment as well. • Patrizio Vanella, Timon Hellwagner, Philipp Deschermeier466 Our paper uses a novel hierarchical approach, which quantifi es the annual differentials in employees subject to social insurance contributions, stratifi ed by gender and country of origin group, for each LMR. The model accounts for demographic and geographic characteristics while including overall economic development. Previous approaches, as illustrated in the literature review, took a shorter-term perspective, which does not suffi ciently refl ect the long-term nature of the labor market integration of refugees. Besides an analysis of the past regional labor market effects of refugee migration, we offer an outlook on how refugee migration might impact the 34 discussed LMRs until 2030 in the form of a deterministic, yet data-driven, projection of age-, sex-, and origin-specifi c refugee immigration. The model also accounts for asylum policy and expected regional demographic developments. Thus, our paper not only provides insights into regionally varying employment effects of refugee migration and the respective interaction with population decline but, from a more methodological point of view, has also shown the general complexity associated with analyzing these effects on regional labor markets in a quantitative and comparative setting. However, inevitably, our empirical approach is subject to several limitations which must be kept in mind when interpreting the results and drawing conclusions. We group and explain these limitations in three categories below: non-capturing statistical risk, small regional case numbers and data availability, as well as other relevant but not explicitly modeled processes. First, refugee migration is highly stochastic and therefore diffi cult to predict, as it is infl uenced by many political, social, economic, and environmental factors. Throughout this paper, we apply a deterministic modeling strategy, since an appropriate stochastic approach should include stochasticity arising from all partial models. Put differently, the nexus of refugee migration and labor market integration – encompassing, inter alia, the initial regional distribution, legal issues connected to labor market integration, individual characteristics, and regional differences in labor demand – is highly complex. Most of those components within partial models are associated with stochasticity, as outlined in Section 3, calling for a coherent approach to address this “cumulative” stochasticity. However, such an approach is not feasible for two reasons. On the one hand, data limitations largely rule out a reliable stochastic modeling strategy (for instance, regional refugee data are rounded, which could lead to strong biases in the variance estimates for small regions). Similar limitations apply to labor market variables as discussed below. Deriving stochastic estimates for our projections with these limited data, based on, e.g., theoretical assumptions, could lead to biased estimates for the projection intervals (either too narrow because of incorrect assumptions or too wide and therefore less meaningful). On the other hand, assuming the availability of the necessary data, a stochastic approach would also exceed the scope of this paper. We do not intend to present a complete forecast approach of regionally varying employment effects due to refugee migration. Rather, given the obvious undercoverage of labor market effects of refugee migration from an explicitly regional perspective in general and in the context of population decline in particular, we propose a novel, though deterministic, approach to addressing these issues for the fi rst time. Keeping these Past and Future Trends in Refugee Migration on the Regional Level in Germany • 467 limitations in mind, a stochastic extension of our approach could be explored in future research. Second, as noted, some major limitations emerge due to data availability. In general, the degree of detail of the refugee statistics is very low. Additionally, small- area data on refugee stocks are inaccurately reported by rounding to account for data privacy. This, however, renders the data inadequate for our statistical analyses. Similarly, our modeling strategy of employment effects exhibits shortcomings. Crucial factors discussed in the literature review, such as education, skills, age, or information on the occupation of refugees, were not included in the analysis due to data availability or complexity. Most of the literature on refugee migration and labor market issues draws on individual-level data to address these factors. In the present paper, using corresponding individual instead of administrative data, such as the comprehensive IAB-BAMF-SOEP survey (see, e.g., Brücker et al. 2019), would mean relying on more detailed information for refugees’ characteristics, but at the same time, and more importantly, on a less appropriate database for comparative macro- level analysis across regions, given the small number of observations across time and the regional coverage in the sampling method. Similarly, relying on detailed administrative data, for example, disaggregated using a classifi cation of occupations, would imply very few or no observations in some gender-country specifi c groups, in particular for small labor market regions. Thus, given the objective of this paper of presenting a novel and rather general analysis of employment effects across regions in the context of population decline, we rely on aggregated numbers of all persons employed and eligible for social security within a region. This acknowledges the limitations stemming from excluding detailed individual characteristics and leaves this shortcoming open to future research. Third, other not explicitly modeled processes, such as the discussed internal migration effects or regionally varying economic conditions, may limit the interpretation of the presented results. As noted in Section 3.6, we do not fully mirror regional employment dynamics. By construction, our model does not allow for leaving employment, no matter whether this might have occurred due to retirement, job loss, death, disability, emigration, voluntarily, or for another reason. Also, regional differences such as sectoral structure, the frequency of job turnovers, or the degree of regional labor market tightness and the respective interaction with demographic structure and individual characteristics of refugees is beyond the scope of our approach, and thus only included implicitly. Importantly, we also only use data on employed persons eligible for social insurance but exclude marginal employment. Possibly regional disproportionally higher or lower integration of refugees into marginal employment is thus not refl ected in our approach. Similarly, by relating projected employment effects to population decline, we implicitly assume that the decrease in the working-age population equals the decrease in labor supply. Given the evidence in the literature concerning regionally varying and, over time, changing labor force participation rates, this limitation must be kept in mind. Moreover, we only model employment effects as additional employees in the context of refugee migration. Thus, spillover effects, e.g., leading to increased employment among natives or other migrants, are not included. Finally, our labor supply analysis • Patrizio Vanella, Timon Hellwagner, Philipp Deschermeier468 excludes Europeans, as described in the methods section of this paper. As the input data includes periods of EU enlargement, the number of countries with free movement of workers has increased over time, leading to declining numbers of asylum seekers from Europe. To avoid biased model specifi cations arising from this, we do not include refugee migration infl ows from European countries in the fi nal parts of this study. 6 Conclusions The labor market integration of refugees in Germany and elsewhere has been discussed extensively in recent years. This development is refl ected in an increasing body of empirical research, mostly from a micro perspective, as illustrated by a comprehensive literature review in this paper. The process of refugees’ labor market integration depends on a multitude of factors. Individually, education and language skills are of crucial importance. Institutionally, bureaucratic hurdles such as employment bans complicate the integration process. Employment effects of refugee migration likely differ across regions, given varying structural and demographic contexts. In particular, this includes varying intensities of projected labor supply decline as well as further intervening processes such as internal migration. Thus, this paper has aimed at investigating whether the region-specifi c interaction of employment dynamics and demographic changes yields regionally varying opportunities for mitigating labor supply decline. By employing a macro- level approach, we believe our paper addresses a substantial shortcoming in the literature, which is strongly focused on micro-level studies. We apply a complex empirical approach. First, as there are no long-term projections of refugee migration, particularly at the regional level, we contribute to the literature by proposing a detailed forecasting strategy. We analyze initial asylum requests by country of origin, gender, and protection rates with the regional distribution according to federal guidelines. Based on developments from 1995 to 2019, we project gender-origin-specifi c refugee migration in German labor market regions until 2030 using principal components analysis. Second, we use past migration values on the one hand and offi cial statistics from the Federal Employment Agency on the other to analyze regional employment effects between 2007 and 2019. That is, we relate the cumulated employment fi gures from our model to both the cumulated refugee infl ow fi gures and the stock of unemployment according to offi cial statistics. In the last step, we combine both the refugee migration projection results and the fi ndings from the econometric model with existing population projections on the regional level and analyze the corresponding interactions. The fi ndings suggest that the employment effects due to refugee migration lead to regionally varying opportunities for mitigating labor supply decline. In particular, for labor market regions surrounding large cities such as Munich and Berlin, employment by refugee migration will contribute more strongly to cushioning the demographically induced declines. Similarly, the results indicate that economically more prosperous regions may expect to experience larger mitigation effects. By Past and Future Trends in Refugee Migration on the Regional Level in Germany • 469 contrast, for most of the labor market regions in Eastern Germany, employment effects of refugee migration are likely to contribute to cushioning population decline effects far less. However, the empirical strategy and the corresponding fi ndings are subject to several limitations, which we outlined in the discussion section of this paper. For example, internal migration after the arrival in Germany, among several other factors, might explain substantial parts of the observed differences. Inevitably, this must be considered in any interpretation of the results. Based upon the introduced modeling approach, the discussed fi ndings, as well as the outlined limitations of this paper, several avenues for future research emerge. First, the paper has introduced a novel modeling approach to investigating the employment effects of refugee migration from a macro perspective, which may be used and improved by other approaches and researchers. Second, aside from methodological improvements such as stochastic approaches, future research is likely to benefi t from better data availability. Individual labor market outcomes of refugees and regional (demographic and economic) development must be understood and analyzed as interdependent processes. Thus, for example, individual-level data coupled with offi cial employment statistics, which is currently not possible, would substantially enhance research results. Third, the observed patterns and corresponding mitigation potentials stem from different, likely interdependent, and in any case interacting processes, such as internal migration, co-ethnic networks, and economic prosperity. Disentangling channels of impact will contribute substantially to our understanding of differing employment and mitigation effects. Fourth, more generally, the fi ndings presented may serve as a starting point for further approaches to investigating past and future economic effects of (refugee) migration in an explicitly regional comparative perspective and across disciplines. Acknowledgements The present study was conducted before the latest intensifi cation of the Russian- Ukrainian confl ict and the availability of data during the pandemic period. Financial support through the joint graduate program in labour market research of the Institute for Employment Research (IAB) and the School of Business and Economics at the University of Erlangen-Nuremberg, GradAB, is gratefully acknowledged by TH. Moreover, we appreciate the helpful comments by the anonymous reviewers of the paper. Authors' Contributions Conceptualization: PV; Methodology: PV and TH; Software: PV and TH; Validation: PV, TH, and PD; Formal Analysis: PV and TH; Investigation: TH, PV, and PD; Data Curation: PV and TH; Writing: PV and TH; Visualization: TH and PV; Supervision: PV; All authors have read and approved the fi nal version of the manuscript. • Patrizio Vanella, Timon Hellwagner, Philipp Deschermeier470 References Aaronson, Stephanie et al. 2014: Labor Force Participation: Recent Developments and Future Prospects. In: Brookings Papers on Economic Activity 45,2: 197-255. https://doi.org/10.1353/eca.2014.0015 Ager, Alastair; Strang, Alison 2008: Understanding Integration: A Conceptual Framework. In: Journal of Refugee Studies 21,2: 166-191. https://doi.org/10.1093/jrs/fen016 Algan, Yann et al. 2010: The Economic Situation of First and Second-Generation Immigrants in France, Germany and the United Kingdom. In: Economic Journal 120,542: 4-30. https://doi.org/10.1111/j.1468-0297.2009.02338.x BAMF 2010: Asyl in Zahlen 2009. Nürnberg: BAMF. BAMF 2011: Das Bundesamt in Zahlen 2010. Asyl, Migration, ausländische Bevölkerung und Intergration. Nürnberg: BAMF. BAMF 2012: Das Bundesamt in Zahlen 2011. Asyl, Migration, ausländische Bevölkerung und Intergration. Nürnberg: BAMF. BAMF 2013: Das Bundesamt in Zahlen 2012. Asyl, Migration und Integration. Nürnberg: BAMF. BAMF 2014: Das Bundesamt in Zahlen 2013. Asyl, Migration und Integration. Nürnberg: BAMF. BAMF 2015: Das Bundesamt in Zahlen 2014. Asyl, Migration und Integration. Nürnberg: BAMF. BAMF 2016: Das Bundesamt in Zahlen 2015. Asyl, Migration und Integration. Nürnberg: BAMF. BAMF 2017: Das Bundesamt in Zahlen 2016. Asyl, Migration und Integration. Nürnberg: BAMF. BAMF 2018: Aktuelle Zahlen zu Asyl. Ausgabe: Dezember 2017. Nürnberg: BAMF. BAMF 2019a: Aktuelle Zahlen zu Asyl. Ausgabe: Dezember 2018. Nürnberg: BAMF. BAMF 2019b: Ablauf des deutschen Asylverfahrens. Ein Überblick über die einzelnen Verfahrensschritte und rechtlichen Grundlagen. Nürnberg: BAMF. BAMF 2019c: Minas. Atlas über Migration, Integration und Asyl. 9. Ausgabe. Nürnberg: BAMF. BAMF 2020a: Aktuelle Zahlen zu Asyl. Ausgabe: Dezember 2019. Nürnberg: BAMF. BAMF 2020b: Migrationsreport 2019. Berlin: BMI. BAMF 2021: Antrags-, Entscheidungs- und Bestandsstatistik 1995-2020. Basilio, Leilanie; Bauer, Thomas K.; Kramer, Anica 2017: Transferability of Human Capital and Immigrant Assimilation: An Analysis for Germany. In: Labour 31,3: 245- 264. https://doi.org/10.1111/labr.12096 BBSR 2021: Raumordnungsprognose nach Kreisen. Bonn: Bundesamt für Bauwesen und Raumordnung. Becker, Sascha O.; Ferrara, Andreas 2019: Consequences of forced migration: A survey of recent fi ndings. In: Labour Economics 59: 1-16. https://doi.org/10.1016/j.labeco.2019.02.007 Beyer, Robert C.M. 2019: Wage Performance of Immigrants in Germany. In: German Economic Review 20,4: 141-169. https://doi.org/10.1111/geer.12159 https://doi.org/10.1353/eca.2014.0015 https://doi.org/10.1093/jrs/fen016 https://doi.org/10.1111/j.1468-0297.2009.02338.x https://doi.org/10.1111/labr.12096 https://doi.org/10.1016/j.labeco.2019.02.007 https://doi.org/10.1111/geer.12159 Past and Future Trends in Refugee Migration on the Regional Level in Germany • 471 Bijak, Jakub et al. 2019: Assessing time series models for forecasting international migration: Lessons from the United Kingdom. In: Journal of Forecasting 38,5: 470- 487. https://doi.org/10.1002/for.2576 BKG 2021: Verwaltungsgebiete 1:1 000 000 (Ebenen), Stand 01.01. (VG1000 01.01.) [https://gdz.bkg.bund.de/index.php/default/digitale-geodaten/verwaltungsgebiete/ verwaltungsgebiete-1-1-000-000-ebenen-stand-01-01-vg1000-ebenen-01-01.html, 19.05.2021]. BMBF 2020: Bericht zum Anerkennungsgesetz 2019. Berlin: Federal Ministry of Education and Research. BMI 2013: 64.539 Asylerstanträge im Jahr 2012 [https://www.bmi.bund.de/SharedDocs/ pressemitteilungen/DE/2013/01/asylzahlen_2012.html, 01.11.2021]. Brell, Courtney; Dustmann, Christian; Preston, Ian 2020: The Labor Market Integration of Refugee Migrants in High-Income Countries. In: Journal of Economic Perspectives 34,1: 94-121. Brenke, Karl; Clemens, Marius 2017: Steigende Erwerbsbeteiligung wird künftig kaum ausreichen, um den demografi schen Wandel in Deutschland zu meistern. DIW Wochenbericht 35: 675-685. Brücker, Herbert; Jaschke, Philipp; Kosyakova, Yuliya 2019: Integrating Refugees into the German Economy and Society: Empirical Evidence and Policy Objectives. Washington, D.C.: Migration Policy Institute. Brücker, Herbert et al. 2019: Zweite Welle der IAB-BAMF-SOEP-Befragung: Gefl üchtete machen Fortschritte bei Sprache und Beschäftigung. IAB-Kurzbericht. Brücker, Herbert; Hauptmann, Andreas; Jaschke, Philipp 2020: Beschränkungen der Wohnortwahl für anerkannte Gefl üchtete: Wohnsitzaufl agen reduzieren die Chancen auf Arbeitsmarktintegration. IAB-Kurzbericht 3/2020. Nürnberg: Institut für Arbeitsmarkt- und Berufsforschung (IAB). Brücker, Herbert et al. 2021: Occupational Recognition and Immigrant Labor Market Outcomes. In: Journal of Labor Economics 39,2: 497-525. https://doi.org/10.1086/710702 Burstein, Ariel et al. 2020: Tradability and the Labor-Market Impact of ImmigrationTradability and the Labor-Market Impact of Immigration: Theory and Evidence From the United States. In: Econometrica 88,3: 1071-1112. https://doi.org/10.3982/ECTA16196 Cai, Lixin 2010: The relationship between health and labour force participation: Evidence from a panel data simultaneous equation model. In: Labour Economics 17,1: 77-90. De La Rica, Sara; Glitz, Albrecht; Ortega, Francesc 2015: Immigration in Europe: Trends, Policies, and Empirical Evidence. In: Chiswick, Barry; Miller, Paul (Eds.): Handbook of the Economics of International Migration, Volume 1B. Oxford/Amsterdam: North- Holland. Deschermeier, Philipp 2011: Population Development of the Rhine-Neckar Metropolitan Area. A Stochastic Population Forecast on the Basis of Functional Data Analysis. In: Comparative Population Studies 36,4: 769-806. https://doi.org/10.4232/10.CPoS-2011-21en Deschermeier, Philipp 2016: Einfl uss der Zuwanderung auf die demografi sche Entwicklung in Deutschland. In: IW-Trends. Vierteljahresschrift zur empirischen Wirtschaftsforschung 43,2: 21-38. https://doi.org/10.2373/1864-810X.16-02-03 Destatis 2020a: Bevölkerung und Erwerbstätigkeit. Schutzsuchende. Ergebnisse des Ausländerzentralregisters. In: Fachserie 1 2.4. Statistisches Bundesamt. https://doi.org/10.1002/for.2576 https://gdz.bkg.bund.de/index.php/default/digitale-geodaten/verwaltungsgebiete/verwaltungsgebiete-1-1-000-000-ebenen-stand-01-01-vg1000-ebenen-01-01.html https://gdz.bkg.bund.de/index.php/default/digitale-geodaten/verwaltungsgebiete/verwaltungsgebiete-1-1-000-000-ebenen-stand-01-01-vg1000-ebenen-01-01.html https://www.bmi.bund.de/SharedDocs/pressemitteilungen/DE/2013/01/asylzahlen_2012.html https://www.bmi.bund.de/SharedDocs/pressemitteilungen/DE/2013/01/asylzahlen_2012.html https://doi.org/10.1086/710702 https://doi.org/10.3982/ECTA16196 https://doi.org/10.4232/10.CPoS-2011-21en https://doi.org/10.2373/1864-810X.16-02-03 • Patrizio Vanella, Timon Hellwagner, Philipp Deschermeier472 Destatis 2020b: Bevölkerung und Erwerbstätigkeit. Erwerbsbeteiligung der Bevölkerung. Ergebnisse des Mikrozensus zum Arbeitsmarkt. In: Fachserie 1 4.1. Statistisches Bundesamt. Dustmann, Christian; Glitz, Albrecht; Vogel, Thorsten 2010: Employment, wages, and the economic cycle: Differences between immigrants and natives. In: European Economic Review 54,1: 1-17. Dustmann, Christian; Frattini, Tommaso 2013: Immigration: The European Experience. In: Card, David; Raphael, Steven (EdS.): Immigration, Poverty and Socioeconomic Inequality. New York: Russel Sage Foundation: 423-456. Dustmann, Christian; Frattini, Tommaso; Preston, Ian P. 2013: The Effect of Immigration along the Distribution of Wages. In: Review of Economic Studies 80,1: 145-173. https://doi.org/10.1093/restud/rds019 Dustmann, Christian et al. 2017: On the economics and politics of refugee migration. In: Economic Policy 32,91: 497-550. https://doi.org/10.1093/epolic/eix008 EASO 2016: Signifi cant Pull/Push Factors for Determining of Asylum-Related Migration. A Literature Review. Luxembourg: Publications Offi ce of the European Union. Eckert, Daniel 2017: Diese Zahlen offenbaren das Ausmaß der Flüchtlingskrise [https:// www.welt.de/wirtschaft/article163753103/Diese-Zahlen-offenbaren-das-Ausmass- der-Fluechtlingskrise.html, 20.05.2021]. efms 2003: Migrationsbericht 2003. Berlin. Engler, Marcus; Schneider, Jan 2015: Aktuelle Entwicklungen in Deutschland [https:// www.bpb.de/gesellschaft/migration/kurzdossiers/207639/asyl-und-asylpolitik, 04.05.2021]. EU 2011: Directive 2011/95/EU of the European Parliament and of the Council of 13 December 2011 on standards for the qualifi cation of third-country nationals or stateless persons as benefi ciaries of international protection, for a uniform status for refugees or for persons eligible for subsidiary protection, and for the content of the protection granted. In: Offi cial Journal of the European Union L 337/9. Eurostat 2021: Glossary: EU enlargements [https://ec.europa.eu/eurostat/statistics- explained/index.php?title=Glossary:EU_enlargements, 01.06.2021]. Fasani, Francesco; Frattini, Tommaso; Minale, Luigi 2018: (The Struggle for) Refugee Integration into the Labour Market: Evidence from Europe. IZA Disscusion Paper. Federal Employment Agency 2021: Statistics of the Federal Employment Agency. Fingleton, Bernard; Garretsen, Harry; Martin, Ron 2012: Recessionary Shocks and Regional Employment: Evidence on the Resilience of U.K. Regions. In: Journal of Regional Science 52,1: 109-133. https://doi.org/10.1111/j.1467-9787.2011.00755.x Fuchs, Johann et al. 2018: Stochastic Forecasting of Labor Supply and Population: An Integrated Model. In: Population research and policy review 37,1: 33-58. https://doi.org/10.1007/s11113-017-9451-3 Fuchs, Johann; Kubis, Alexander; Schneider, Lutz 2019: Zuwanderung und Digitalisierung. Wie viel Migration aus Drittstaaten benötigt der deutsche Arbeitsmarkt künftig? Gütersloh. https://doi.org/10.11586/2019013 Galantino, Maria G. 2022: The migration-terrorism nexus: An analysis of German and Italian press coverage of the 'refugee crisis'. In: European Journal of Criminology 19,2: 259-281. https://doi.org/10.1177/1477370819896213 Gehrsitz, Markus; Ungerer, Martin 2017: Jobs , Crime , and Votes : A Short-run Evaluation of the Refugee Crisis in Germany. IZA Disscusion Paper. https://doi.org/10.1093/restud/rds019 https://doi.org/10.1093/epolic/eix008 https://www.welt.de/wirtschaft/article163753103/Diese-Zahlen-offenbaren-das-Ausmass-der-Fluechtlingskrise.html https://www.welt.de/wirtschaft/article163753103/Diese-Zahlen-offenbaren-das-Ausmass-der-Fluechtlingskrise.html https://www.welt.de/wirtschaft/article163753103/Diese-Zahlen-offenbaren-das-Ausmass-der-Fluechtlingskrise.html https://www.welt.de/wirtschaft/article163753103/Diese-Zahlen-offenbaren-das-Ausmass-der-Fluechtlingskrise.html https://www.bpb.de/gesellschaft/migration/kurzdossiers/207639/asyl-und-asylpolitik https://www.bpb.de/gesellschaft/migration/kurzdossiers/207639/asyl-und-asylpolitik https://ec.europa.eu/eurostat/statistics-explained/index.php?title=Glossary:EU_enlargements https://ec.europa.eu/eurostat/statistics-explained/index.php?title=Glossary:EU_enlargements https://ec.europa.eu/eurostat/statistics-explained/index.php?title=Glossary:EU_enlargements https://doi.org/10.1111/j.1467-9787.2011.00755.x https://doi.org/10.1007/s11113-017-9451-3 https://doi.org/10.11586/2019013 https://doi.org/10.1177/1477370819896213 Past and Future Trends in Refugee Migration on the Regional Level in Germany • 473 Geis, Wido; Orth, Anja K. 2016: Flüchtlinge regional besser verteilen. Ausgangslage und Ansatzpunkte für einen neuen Verteilungsmechanismus. Köln: Robert Bosch Stiftung. GENESIS-Online 2021: Bevölkerung: Bundesländer, Stichtag, Altersjahre, Nationalität, Geschlecht, Familienstand. Fortschreibung des Bevölkerungsstandes [https://www- genesis.destatis.de/genesis/online, 29.04.2021]. GKW 2021a: Königsteiner Schlüssel für das Jahr 2010-2018 [https://www.gwk-bonn. de/fi leadmin/Redaktion/Dokumente/Papers/koenigsteiner-schluessel-2010-2018.pdf, 16.02.2021]. GWK 2021b: Übersicht Königsteiner Schlüssel ab 1995. Bonn: Gemeinsame Wissenschaftskonferenz. Guichard, Lucas 2020: Self-selection of Asylum Seekers: Evidence From Germany. In: Demography 57,3: 1089-1116. https://doi.org/10.1007/s13524-020-00873-9 Gustafsson, Bjorn A.; MacInnes, Hanna; Österberg, Torun 2017: Age at immigration matters for labor market integration – the Swedish example. In: IZA Journal of Development and Migration 7,1. https://doi.org/10.1186/s40176-017-0087-1 Haase, Marianne; Jugl, Jan C. 2007: Asyl- und Flüchtlingspolitik der EU [https://www.bpb. de/gesellschaft/migration/dossier-migration-ALT/56551/asyl-fl uechtlingspolitik?p=0, 04.05.2021]. Hammond, Laura 2014: History, overview, trends and issues in major Somali refugee displacements in the near region. In: New Issues in Refugee Research 2014,268. Heidelberg Institute for International Confl ict Research 2007: Konfl iktbarometer 2007. Krisen – Kriege – Putsche. Verhandlungen – Interventionen – Friedensschlüsse. Heidelberg: Heidelberg Institute for International Confl ict Research. Heidelberg Institute for International Confl ict Research 2009: Confl ict Barometer 2009. Crises – Wars – Coups d'État. Negotiations – Mediations – Peace Settlements. Heidelberg: Heidelberg Institute for International Confl ict Research. Heidelberg Institute for International Confl ict Research 2010: Confl ict Barometer 2010. Crises––Wars – Coups d'État. Negotiations – Mediations – Peace Settlements. Heidelberg: Heidelberg Institute for International Confl ict Research. Heidelberg Institute for International Confl ict Research 2011: Confl ict Barometer 2011. disputes – non-violent crises – violent crises – limited wars – wars. Heidelberg: Heidelberg Institute for In-ternational Confl ict Research. Heidelberg Institute for International Confl ict Research 2014: Confl ict Barometer 2013. disputes – non-violent crises – violent crises – limited wars – wars. Heidelberg: Heidelberg Institute for In-ternational Confl ict Research. Heidelberg Institute for International Confl ict Research 2016: Confl ict Barometer 2015. disputes – non-violent crises – violent crises – limited wars – wars. Heidelberg: Heidelberg Institute for In-ternational Confl ict Research. Heidelberg Institute for International Confl ict Research 2018: Confl ict Barometer 2018. disputes – non-violent crises – violent crises – limited wars – wars. Heidelberg: Heidelberg Institute for In-ternational Confl ict Research. Heidelberg Institute for International Confl ict Research 2019: Confl ict Barometer 2019. disputes – non-violent crises – violent crises – limited wars – wars. Heidelberg: Heidelberg Institute for In-ternational Confl ict Research. Human Rights Watch 2010: Iran: Crisis Deepening One Year After Disputed Elections [https://www.refworld.org/docid/4c15ef001e.html, 10.03.2021]. https://www-genesis.destatis.de/genesis/online https://www-genesis.destatis.de/genesis/online https://www-genesis.destatis.de/genesis/online https://www.gwk-bonn.de/fileadmin/Redaktion/Dokumente/Papers/koenigsteiner-schluessel-2010-2018.pdf https://www.gwk-bonn.de/fileadmin/Redaktion/Dokumente/Papers/koenigsteiner-schluessel-2010-2018.pdf https://doi.org/10.1007/s13524-020-00873-9 https://doi.org/10.1186/s40176-017-0087-1 https://www.bpb.de/gesellschaft/migration/dossier-migration-ALT/56551/asyl-fluechtlingspolitik?p=0 https://www.bpb.de/gesellschaft/migration/dossier-migration-ALT/56551/asyl-fluechtlingspolitik?p=0 https://www.refworld.org/docid/4c15ef001e.html • Patrizio Vanella, Timon Hellwagner, Philipp Deschermeier474 Hunkler, Christian; Edele, Aileen; Schipolowski, Stefan 2021: Die Bedeutung von Bildungsressourcen für die Arbeitsmarktintegration Gefl üchteter am Beispiel syrischer Asylsuchender in Deutschland. In: Journal for Educational Research Online 13,1: 157- 181. https://doi.org/10.31244/jero.2021.01.06 International Centre for Migration Policy Development 2015: The State of Migration in Georgia. Report developed in the framework of the EU-funded Enhancing Georgia’s Migration Management (ENIGMMA) project. Vienna: International Centre for Migration Policy Development. Kropp, Per; Schwengler, Barbara 2011: Abgrenzung von Arbeitsmarktregionen – ein Methodenvorschlag. In: Raumforschung und Raumordnung 69,1: 45-62. https://doi.org/10.1007/s13147-011-0076-4 Kropp, Per; Schwengler, Barbara 2016: Three-Step Method for Delineating Functional Labour Market Regions. In: Regional Studies 50,3: 429-445. https://doi.org/10.1080/00343404.2014.923093 Marbach, Moritz; Hainmueller, Jens; Hangartner, Dominik 2018: The long-term impact of employment bans on the economic integration of refugees. In: Science Advances 4,9: 1-7. https://doi.org/10.1126/sciadv.aap9519 Maretzke, Steffen et al. 2021: Raumordnungsprognose 2040. Bevölkerungsprognose: Entwicklung nach Altersgruppen. BBSR-Analysen KOMPAKT. Rasche, Ute 2012: Ein Urteil und seine Folgen [https://www.faz.net/aktuell/politik/inland/ asylbewerber-ein-urteil-und-seine-folgen-11926946.html, 01.06.2021]. Rösch, Tabea et al. 2020: Integration von Gefl üchteten in ländlichen Räumen. Forschungsbericht 36. Nürnberg: Bundesamt für Migration und Flüchtlinge. Ruiz, Isabel; Vargas-Silva, Carlos 2018: Differences in labour market outcomes between natives, refugees and other migrants in the UK. In: Journal of Economic Geography 18,4: 855-885. https://doi.org/10.1093/jeg/lby027 Schock, Katrin et al. 2016: Impact of new traumatic or stressful life events on pre-existing PTSD in traumatized refugees: Results of a longitudinal study. In: European Journal of Psychotraumatology 7,1. https://doi.org/10.3402/ejpt.v7.32106 Schulte, Hubert 2015: Hamburg, Hessen und Nordrhein-Westfalen: drei klassische Zahlerländer im Abwärtstrend? In: Wirtschaftsdienst. Zeitschrift für Wirtschaftspolitik 95,7: 476-481. https://doi.org/10.1007/s10273-015-1850-7 Sohst, Rhea R. et al. 2020: The Future of Migration to Europe: A Systematic Review of the Literature on Migration Scenarios and Forecasts. Geneva: International Organization for Migration. Statistische Ämter des Bundes und der Länder 2021: Bevölkerung nach Geschlecht – Stichtag 31.12. – regionale Ebenen. Fortschreibung des Bevölkerungsstandes [https:// www.regionalstatistik.de/genesis/online/, 16.02.2021]. SVR Integration und Migration 2018: Fakten zur Asylpolitik 2017 [https:// f luechtlingsnetzwerk.de/downloads/infomat/svr_fakten_zur_asylpolitik1.pdf, 01.06.2021]. UNHCR 2019: UNHCR Projected Global Resettlement Needs 2020. Geneva: UNHCR. UNHCR 2020: Mid-Year Trends Report 2020. Geneva: UNHCR. Vanella, Patrizio 2018: Stochastic Forecasting of Demographic Components Based on Principal Component Analyses. In: Athens Journal of Sciences 5,3: 223-246. https://doi.org/10.30958/ajs_v5i3 https://doi.org/10.31244/jero.2021.01.06 https://doi.org/10.1007/s13147-011-0076-4 https://doi.org/10.1080/00343404.2014.923093 https://doi.org/10.1126/sciadv.aap9519 https://www.faz.net/aktuell/politik/inland/asylbewerber-ein-urteil-und-seine-folgen-11926946.html https://www.faz.net/aktuell/politik/inland/asylbewerber-ein-urteil-und-seine-folgen-11926946.html https://doi.org/10.1093/jeg/lby027 https://doi.org/10.3402/ejpt.v7.32106 https://doi.org/10.1007/s10273-015-1850-7 https://www.regionalstatistik.de/genesis/online/ https://www.regionalstatistik.de/genesis/online/ https://fluechtlingsnetzwerk.de/downloads/infomat/svr_fakten_zur_asylpolitik1.pdf https://fluechtlingsnetzwerk.de/downloads/infomat/svr_fakten_zur_asylpolitik1.pdf https://doi.org/10.30958/ajs_v5i3 Past and Future Trends in Refugee Migration on the Regional Level in Germany • 475 Vanella, Patrizio; Deschermeier, Philipp 2018: A stochastic Forecasting Model of international Migration in Germany. In: Kapella, Olaf; Schneider, Norbert F.; Rost, Harald (Eds.): Familie – Bildung – Migration. Familienforschung im Spannungsfeld zwischen Wissenschaft, Politik und Praxis. Tagungsband zum 5. Europäischen Fachkongress Familienforschung. Opladen/Berlin/Toronto: Verlag Barbara Budrich: 261-280. Vanella, Patrizio; Deschermeier, Philipp 2019: A Principal Component Simulation of Age-Specifi c Fertility – Impacts of Family and Social Policy on Reproductive Behavior in Germany. In: Population Review 58,1: 78-109. https://doi.org/10.1353/prv.2019.0002 Vanella, Patrizio; Deschermeier, Philipp 2020: A Probabilistic Cohort-Component Model for Population Forecasting. The Case of Germany. In: Journal of Population Ageing 13,4: 513-545. https://doi.org/10.1007/s12062-019-09258-2 Vanella, Patrizio; Deschermeier, Philipp; Wilke, Christina B. 2020: An Overview of Population Projections – Methodological Concepts, International Data Availability, and Use Cases. In: Forecasting 2,3: 346-363. https://doi.org/10.3390/forecast2030019 VGRdL 2019: Bruttoinlandsprodukt, Bruttowertschöpfung in den kreisfreien Städten und Landkreisen der Bundesrepublik Deutschland 1992 und 1994 bis 2018 [https://www.statistikportal.de/de/veroeffentlichungen/bruttoinlandsprodukt- bruttowertschoepfung-0, 01.06.2021]. Wanger, Susanne 2020: Entwicklung von Erwerbstätigkeit, Arbeitszeit und Arbeitsvolumen nach Geschlecht. Ergebnisse der IAB-Arbeitszeitrechnung nach Alter und Geschlecht (AZR AG) für die Jahre 1991 - 2019. In: IAB-Forschungsbericht 16/2020. Zeit 2019: Gedenken an Putschversuch vor drei Jahren [https://www.zeit.de/ politik/ausland/2019-07/tuerkei-putschversuch-recep-tayyip-erdogan-jahrestag- gedenkfeier?utm_referrer=https%3A%2F%2Fde.wikipedia.org%2F#!top-of- overscroll, 01.06.2021]. Zika, Gerd et al. 2020: Langfristige Folgen von Demografi e und Strukturwandel für regionale Arbeitsmärkte. In: IAB-Forschungsbericht 1-2020. Date of submission: 22.06.2021 Date of acceptance: 01.11.2022 Dr. Patrizio Vanella ().* aQua Institute, Health Reporting & Biometrics Department. Göttingen, Germany. University of Rostock, Chair of Empirical Methods in Social Science and Demography. Rostock, Germany. E-mail: Patrizio.Vanella@Aqua-Institut.de URL: https://www.researchgate.net/profi le/Patrizio-Vanella Timon Hellwagner.* GradAB doctoral program and Department of Forecasts and Macroeconomic Analyses (MAKRO), Institute for Employment Research (IAB) of the Federal Employment Agency (BA). Nürnberg, Germany. E-mail: timon.hellwagner@iab.de URL: https://orcid.org/0000-0001-7480-3437 Dr. Philipp Deschermeier. German Economic Institute (IW), Global and Regional Markets Unit. Köln, Germany. E-mail: deschermeier@iwkoeln.de URL: h ps://www.iwkoeln.de/en/ins tute/who-we-are/philipp-deschermeier.html * Equal contribu on https://doi.org/10.1353/prv.2019.0002 https://doi.org/10.1007/s12062-019-09258-2 https://doi.org/10.3390/forecast2030019 https://www.statistikportal.de/de/veroeffentlichungen/bruttoinlandsprodukt-bruttowertschoepfung-0 https://www.statistikportal.de/de/veroeffentlichungen/bruttoinlandsprodukt-bruttowertschoepfung-0 https://www.statistikportal.de/de/veroeffentlichungen/bruttoinlandsprodukt-bruttowertschoepfung-0 https://www.zeit.de/politik/ausland/2019-07/tuerkei-putschversuch-recep-tayyip-erdogan-jahrestag-gedenkfeier?utm_referrer=https%3A%2F%2Fde.wikipedia.org%2F#!top-of-overscroll https://www.zeit.de/politik/ausland/2019-07/tuerkei-putschversuch-recep-tayyip-erdogan-jahrestag-gedenkfeier?utm_referrer=https%3A%2F%2Fde.wikipedia.org%2F#!top-of-overscroll https://www.zeit.de/politik/ausland/2019-07/tuerkei-putschversuch-recep-tayyip-erdogan-jahrestag-gedenkfeier?utm_referrer=https%3A%2F%2Fde.wikipedia.org%2F#!top-of-overscroll https://www.zeit.de/politik/ausland/2019-07/tuerkei-putschversuch-recep-tayyip-erdogan-jahrestag-gedenkfeier?utm_referrer=https%3A%2F%2Fde.wikipedia.org%2F#!top-of-overscroll https://www.zeit.de/politik/ausland/2019-07/tuerkei-putschversuch-recep-tayyip-erdogan-jahrestag-gedenkfeier?utm_referrer=https%3A%2F%2Fde.wikipedia.org%2F#!top-of-overscroll mailto:Patrizio.Vanella@Aqua-Institut.de https://www.researchgate.net/profi mailto:timon.hellwagner@iab.de https://orcid.org/0000-0001-7480-3437 mailto:deschermeier@iwkoeln.de http://www.iwkoeln.de/en/ins%ED%AF%80%ED%B6%9F • Patrizio Vanella, Timon Hellwagner, Philipp Deschermeier476 Y ea r La rg es t a b so lu te (r el at iv e) O ri g in C au se s in cr ea se w .r .t . p re vi o u s ye ar 20 07 2, 12 0 (1 04 % ) Ir aq Fu rt h er e sc al at io n o f v ar io u s co n fl i ct s as a c o n se q u en ce o f t h e Ir aq W ar 20 08 2, 50 9 (5 8% ) Ir aq S ee a b o ve 20 09 2, 71 8 (4 14 % ) A fg h an is ta n A rm ed c o n fl i ct s, t er ro r at ta ck s b y Ta lib an d u e to p re si d en ti al e le ct io n 20 10 4, 39 7 (7 57 % ) S er b ia A b ili ty t o le g al ly c ro ss E U b o rd er s si n ce a u tu m n 20 09 , a ft er m at h o f fi n an ci al c ri si s 2, 53 0 (7 5% ) A fg h an is ta n Fu rt h er e sc al at io n o f p re vi o u s co n fl i ct s 2, 35 7 (2 ,1 62 % ) N o rt h M ac ed o n ia A b ili ty t o le g al ly c ro ss E U b o rd er s si n ce a u tu m n 20 09 , a ft er m at h o f fi n an ci al c ri si s 1, 88 9 (5 46 % ) S o m al ia V io le n ce o f A l S h ab aa b a g ai n st c it iz en s, d ro u g h ts , e co n o m ic d o w n tu rn 1, 30 5 (1 12 % ) Ir an E m ig ra ti o n af te r vi o le n t s u p p re ss io n o f o p p o si ti o n al d em o n st ra ti o n s ag ai n st d is p u te d p re si d en ti al e le ct io n o f A h m ad in ej ad in 2 00 9 20 11 1, 86 2 (3 2% ) A fg h an is ta n S ee a b o ve 1, 69 9 (2 02 % ) P ak is ta n V io le n t c o n fl i ct s b et w ee n va ri o u s re lig io u s g ro u p s an d t h e m ili ta ry 1, 14 4 (7 7% ) S yr ia A rm ed c o n fl i ct s in t h e af te rm at h o f t h e “A ra b S p ri n g ” 20 12 3, 89 8 (8 5% ) S er b ia M ig ra ti o n af te r fi n an ci al c ri si s, t ri g g er ed b y h ig h in fl o w s fr o m M id d le E as te rn c o u n tr ie s 3, 56 7 (1 35 % ) S yr ia O n g o in g co n fl i ct s af te r 20 11 3, 41 5 (3 02 % ) N o rt h M ac ed o n ia S ee S er b ia 1, 72 0 (5 64 % ) B o sn ia a n d H er ze g o vi n a S ee S er b ia 1, 51 3 (9 0% ) R u ss ia O p p re ss io n o f o p p o si ti o n al d em o n st ra to rs b y g o ve rn m en t f o rc es a ft er th e th ir d p re si d en ti al e le ct io n o f P u ti n A p p en d ix Ta b . A 1: M aj o r an nu al c ha ng es a nd c au se s in in it ia l a sy lu m r eq ue st s in G er m an y b y o ri g in o f r ef ug ee s Past and Future Trends in Refugee Migration on the Regional Level in Germany • 477 Ta b . A 1: C o n tin ua tio n Y ea r La rg es t a b so lu te (r el at iv e) O ri g in C au se s in cr ea se w .r .t . p re vi o u s ye ar 20 13 20 ,3 05 ( 90 % ) C en tr al a n d E as t E u ro p e O n g o in g “s n o w b al l e ff ec t” 14 ,0 88 (1 69 % ) A fr ic a O n g o in g “s n o w b al l e ff ec t” , i n cr ea se d a g g re ss io n b y Is la m ic t er ro r g ro u p s in t h e af te rm at h o f t h e “A ra b S p ri n g ” 9, 58 6 (2 9% ) A si a O n g o in g “s n o w b al l e ff ec t” , i n cr ea se d a g g re ss io n b y Is la m ic t er ro r g ro u p s in th e af te rm at h o f t h e “A ra b S p ri n g ” 1, 04 1 (1 79 % ) U n kn o w n o r n o c it iz en sh ip O n g o in g “s n o w b al l e ff ec t” , ( in s o m e ca se s p u rp o se ly ) l o st id en ti fi ca ti o n d o cu m en ts 20 14 32 ,8 65 ( 77 % ) A si a E m er g in g o f I sl am ic S ta te (I S ) i n S yr ia a n d Ir aq 16 ,9 07 ( 75 % ) A fr ic a O n g o in g “s n o w b al l e ff ec t” , i n cr ea se d a g g re ss io n b y Is la m ic t er ro r g ro u p s, e .g . B o ko H ar am 10 ,5 18 (1 25 % ) C en tr al a n d E as t E u ro p e O n g o in g “s n o w b al l e ff ec t” 3, 19 1 (1 97 % ) U n kn o w n o r n o c it iz en sh ip O n g o in g “s n o w b al l e ff ec t” , ( in s o m e ca se s p u rp o se ly ) l o st id en ti fi ca ti o n d o cu m en ts 20 15 17 4, 77 8 (2 32 % ) A si a In cr ea se d a g g re ss io n b y va ri o u s Is la m ic t er ro r g ro u p s 80 ,7 95 (1 51 % ) C en tr al a n d E as t E u ro p e O n g o in g “s n o w b al l e ff ec t” 10 ,8 30 (2 25 % ) U n kn o w n o r n o c it iz en sh ip O n g o in g “s n o w b al l e ff ec t” , ( in s o m e ca se s p u rp o se ly ) l o st id en ti fi ca ti o n d o cu m en ts 1, 28 3 (3 3% ) N ig er ia In cr ea se d a g g re ss io n b y B o ko H ar am • Patrizio Vanella, Timon Hellwagner, Philipp Deschermeier478 Ta b . A 1: C o n tin ua tio n Y ea r La rg es t a b so lu te (r el at iv e) O ri g in C au se s in cr ea se w .r .t . p re vi o u s ye ar 20 16 31 4, 27 2 (1 26 % ) A si a In cr ea se d a g g re ss io n b y va ri o u s Is la m ic t er ro r g ro u p s, d el ay ed p ro ce ss in g o f a sy lu m r eq u es ts 38 ,5 04 ( 92 % ) A fr ic a O n g o in g “s n o w b al l e ff ec t” , d el ay ed p ro ce ss in g o f a sy lu m r eq u es ts 5, 72 8 (1 09 % ) R u ss ia E sc al at in g co n fl i ct b et w ee n g o ve rn m en t a n d o p p o si ti o n, t h e la tt er in cr ea si n g ly fl ee in g th e co u n tr y 4, 84 3 (3 1% ) U n kn o w n o r n o c it iz en sh ip D el ay ed p ro ce ss in g o f a sy lu m r eq u es ts 3, 88 3 (2 59 % ) Tü rk iy e Fl ig h t o f o p p o si ti o n af te r in cr ea se d s u p p re ss io n b y th e g o ve rn m en t i n th e af te rm at h o f f ai le d m ili ta ry c o u p, in cr ea se d p re ss u re o n su sp ec te d IS c o lla b o ra to rs 20 17 2, 64 4 (4 9% ) Tü rk iy e O n g o in g em ig ra ti o n b y o p p o si ti o n an d IS s u sp ec ts 20 18 2, 35 7 (3 0% ) N ig er ia O n g o in g co n fl i ct s w it h Is la m ic t er ro r g ro u p s 2, 24 9 (2 6% ) Ir an In cr ea se d p ro te st s ag ai n st t h e p o lit ic al s ys te m 2, 13 3 (2 7% ) Tü rk iy e O n g o in g em ig ra ti o n b y o p p o si ti o n an d IS s u sp ec ts S o ur ce : B A M F 20 20 b ; B M I 20 13 ; H am m on d 20 14 ; H ei de lb er g In st itu te f or I nt er na tio na l C on fl i ct R es ea rc h 20 07 , 20 09 , 20 10 , 20 11 , 20 14 , 20 18 , 20 19 ; H um an R ig ht s W at ch 2 01 0; I nt er na tio na l C en tr e fo r M ig ra tio n Po lic y D ev el op m en t 20 15 ; R as ch e 20 12 ; Va ne lla /D es ch er m ei er 2 01 8. Past and Future Trends in Refugee Migration on the Regional Level in Germany • 479 Tab. A2: Distribution of annual initial asylum requests by age and sex [in %] i Sex Age 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 1 Male 0-15 14.60 16.41 17.71 17.71 16.68 14.9 14.59 16.68 20.75 22.95 24.18 2 Male 16-17 4.81 4.84 4.30 3.50 2.67 2.87 3.68 4.71 4.35 2.95 2.37 3 Male 18-24 19.28 16.20 15.62 13.74 15.70 18.07 19.94 17.86 13.62 10.08 9.05 4 Male 25-29 12.00 10.60 10.81 9.75 10.72 11.38 11.50 9.96 7.55 6.48 6.12 5 Male 30-34 6.62 6.51 6.66 6.73 7.07 7.54 7.45 6.33 5.21 4.98 5.08 6 Male 35-39 4.01 3.78 3.96 4.13 4.32 4.75 4.80 4.03 3.45 3.45 3.52 7 Male 40-44 2.04 1.95 2.25 2.53 2.58 2.90 3.10 2.46 2.15 2.21 2.43 8 Male 45-49 1.16 1.17 1.32 1.42 1.52 1.75 1.94 1.61 1.40 1.52 1.54 9 Male 50-54 0.56 0.64 0.74 0.94 0.93 1.06 1.07 0.92 0.85 0.96 0.95 10 Male 55-59 0.43 0.44 0.48 0.57 0.55 0.65 0.54 0.54 0.52 0.54 0.62 11 Male 60-64 0.19 0.31 0.30 0.36 0.31 0.35 0.29 0.33 0.33 0.34 0.38 12 Male 65+ 0.33 0.40 0.37 0.39 0.34 0.33 0.25 0.26 0.28 0.26 0.26 13 Female 0-15 12.98 14.61 14.11 15.24 14.95 13.05 11.89 13.64 18.63 21.18 22.33 14 Female 16-17 1.46 1.54 1.54 1.33 1.12 0.95 0.95 1.16 1.26 1.27 1.24 15 Female 18-24 6.15 6.15 5.71 5.33 5.29 5.24 4.88 5.65 5.23 5.15 4.6 16 Female 25-29 4.49 4.53 4.61 4.65 4.42 4.06 3.72 4.10 3.81 4.00 3.89 17 Female 30-34 3.27 3.56 3.59 3.77 3.64 3.31 3.12 3.29 3.41 3.70 3.54 18 Female 35-39 1.95 2.14 2.32 2.61 2.48 2.22 2.27 2.27 2.53 2.87 2.83 19 Female 40-44 1.07 1.48 1.47 1.66 1.57 1.50 1.47 1.44 1.67 1.92 1.89 20 Female 45-49 0.82 0.87 1.01 1.17 1.01 1.02 0.97 1.02 1.10 1.28 1.20 21 Female 50-54 0.54 0.59 0.77 0.83 0.81 0.76 0.63 0.69 0.70 0.82 0.85 22 Female 55-59 0.43 0.51 0.61 0.67 0.58 0.57 0.42 0.45 0.51 0.50 0.51 23 Female 60-64 0.35 0.31 0.47 0.43 0.34 0.36 0.25 0.29 0.33 0.28 0.31 24 Female 65+ 0.45 0.45 0.56 0.54 0.41 0.42 0.29 0.31 0.36 0.30 0.30 Source: BAMF 2010, 2011, 2012, 2013, 2014, 2015, 2016, 2017, 2018, 2019a, 2020a; authors’ calculation and design. • Patrizio Vanella, Timon Hellwagner, Philipp Deschermeier480 N u m b er N am e D is tr ic ts 1 1 H am b u rg Fl en sb u rg , K ie l, Lü b ec k, N eu m ü n st er , L K D it h m ar sc h en , L K H er zo g tu m L au en b u rg , L K N o rd fr ie sl an d , LK O st h o ls te in , L K P in n eb er g , L K P lö n, L K R en d sb u rg -E ck er n fö rd e, L K S ch le sw ig -F le n sb u rg , LK S eg eb er g , L K S te in b u rg , L K S to rm ar n, H am b u rg , L K H ar b u rg , L K L ü ch o w -D an n en b er g , L K L ü n eb u rg , LK S ta d e, L K U el ze n, S ch w er in , L K N o rd w es tm ec kl en b u rg , L K L u d w ig sl u st -P ar ch im 2 G ö tt in g en LK N o rt h ei m , L K G ö tt in g en , L K E ic h sf el d , L K N o rd h au se n 3 H an n o ve r B ra u n sc h w ei g , S al zg it te r, W o lf sb u rg , L K G if h o rn , L K G o sl ar , L K H el m st ed t, L K P ei n e, L K W o lf en b ü tt el , LK R eg io n H an n o ve r, LK H am el n -P yr m o n t, L K H ild es h ei m , L K N ie n b u rg ( W es er ), LK S ch au m b u rg , LK C el le , L K H ei d ek re is 4 B re m en LK D ie p h o lz , L K C u xh av en , L K O st er h o lz , L K R o te n b u rg ( W ü m m e) , L K V er d en , D el m en h o rs t, E m d en , O ld en b u rg , W ilh el m sh av en , L K A m m er la n d , L K A u ri ch , L K C lo p p en b u rg , L K F ri es la n d , L K L ee r, LK O ld en b u rg , L K W es er m ar sc h, L K W it tm u n d , B re m en , B re m er h av en 5 D ü ss el d o rf /R u h r D ü ss el d o rf , D u is b u rg , E ss en , K re fe ld , M ö n ch en g la d b ac h, M ü lh ei m a n d er R u h r, O b er h au se n, R em sc h ei d , S o lin g en , W u p p er ta l, LK K le ve , L K M et tm an n, L K R h ei n -K re is N eu ss , L K V ie rs en , L K W es el , LK H ei n sb er g , B o tt ro p , G el se n ki rc h en , L K R ec kl in g h au se n, B o ch u m , D o rt m u n d , H ag en , H am m , H er n e, LK E n n ep e- R u h r- K re is , L K H o ch sa u er la n d kr ei s, L K M är ki sc h er K re is , L K S o es t, L K U n n a 6 K ö ln B o n n, K ö ln , L ev er ku se n, L K S tä d te re g io n A ac h en , L K D ü re n, L K R h ei n -E rf t- K re is , L K E u sk ir ch en , LK O b er b er g is ch er K re is , L K R h ei n is ch -B er g is ch er K re is , L K R h ei n -S ie g K re is K o b le n z, L K A h rw ei le r, LK C o ch em -Z el l, LK M ay en -K o b le n z, L K N eu w ie d , L K R h ei n -H u n sr ü ck -K re is , L K W es te rw al d kr ei s 7 M ü n st er /O sn ab rü ck O sn ab rü ck , L K E m sl an d , L K G ra fs ch af t B en th ei m , L K O sn ab rü ck , L K V ec h ta , M ü n st er , L K B o rk en , LK C o es fe ld , L K S te in fu rt , L K W ar en d o rf 8 B ie le fe ld /P ad er b o rn LK H o lz m in d en , B ie le fe ld , L K G ü te rs lo h, L K H er fo rd , L K H ö xt er , L K L ip p e, L K M in d en -L ü b b ec ke , LK P ad er b o rn 9 S ie g en LK O lp e, L K S ie g en -W it tg en st ei n, L K A lt en ki rc h en ( W es te rw al d ) 10 Fr an kf u rt D ar m st ad t, F ra n kf u rt (M ai n) , O ff en b ac h (M ai n) , W ie sb ad en , L K D ar m st ad t- D ie b u rg , L K G ro ß -G er au , LK H o ch ta u n u sk re is , L K M ai n -K in zi g -K re is , L K M ai n -T au n u s- K re is , L K O d en w al d kr ei s, L K O ff en b ac h, LK R h ei n g au -T au n u s- K re is , L K W et te ra u kr ei s, L K G ie ß en , L K L ah n -D ill -K re is , L K L im b u rg -W ei lb u rg , LK M ar b u rg -B ie d en ko p f, LK V o g el sb er g kr ei s, L K F u ld a, L K B ad K re u zn ac h, L K B ir ke n fe ld , L K R h ei n -L ah n - K re is , M ai n z, L K A lz ey -W o rm s, L K M ai n z- B in g en , A sc h af fe n b u rg , L K A sc h af fe n b u rg , L K M ilt en b er g Ta b . A 3: A ss ig nm en t o f d is tr ic ts t o la b o r m ar ke t r eg io n s 1 LK s ta nd s fo r La nd kr ei se , w hi ch a re d is tr ic ts c o m b in ed fr o m d iff er en t ( m o st ly ) s m al le r m un ic ip al iti es . Past and Future Trends in Refugee Migration on the Regional Level in Germany • 481 Ta b . A 3: C o n tin ua tio n N u m b er N am e D is tr ic ts 1 11 K as se l K as se l, LK H er sf el d -R o te n b u rg , L K K as se l, LK S ch w al m -E d er -K re is , L K W al d ec k- Fr an ke n b er g , LK W er ra -M ei ß n er -K re is 12 Tr ie r Tr ie r, LK B er n ka st el -W it tl ic h, L K E if el kr ei s B it b u rg -P rü m , L K V u lk an ei fe l, LK T ri er -S aa rb u rg 13 S tu tt g ar t S tu tt g ar t, L K B ö b lin g en , L K E ss lin g en , L K G ö p p in g en , L K L u d w ig sb u rg , L K R em s- M u rr -K re is , H ei lb ro n n, LK H ei lb ro n n, L K H o h en lo h ek re is , L K S ch w äb is ch H al l, LK M ai n -T au b er -K re is , L K H ei d en h ei m , LK O st al b kr ei s, L K N ec ka r- O d en w al d -K re is , P fo rz h ei m , L K C al w , L K E n zk re is , L K F re u d en st ad t, LK R eu tl in g en , L K T ü b in g en , L K Z o lle rn al b kr ei s 14 K ar ls ru h e/ M an n h ei m LK B er g st ra ß e, F ra n ke n th al (P fa lz ), La n d au in d er P fa lz , L u d w ig sh af en a m R h ei n, N eu st ad t a n d er W ei n st ra ß e, S p ey er , W o rm s, L K B ad D ü rk h ei m , L K D o n n er sb er g kr ei s, L K G er m er sh ei m , L K S ü d lic h e W ei n st ra ß e, L K R h ei n -P fa lz -K re is , B ad en -B ad en , K ar ls ru h e, L K K ar ls ru h e, L K R as ta tt , H ei d el b er g , M an n h ei m , L K R h ei n -N ec ka r- K re is 15 Fr ei b u rg Fr ei b u rg (B re is g au ), LK B re is g au -H o ch sc h w ar zw al d , L K E m m en d in g en , L K O rt en au kr ei s, L K L ö rr ac h, LK W al d sh u t 16 V ill in g en -S ch w en n in g en LK R o tt w ei l, LK S ch w ar zw al d -B aa r- K re is , L K T u tt lin g en 17 K o n st an z LK K o n st an z 18 U lm U lm , L K A lb -D o n au -K re is , L K B ib er ac h, L K S ig m ar in g en , L K N eu -U lm 19 R av en sb u rg LK B o d en se ek re is , L K R av en sb u rg , L K L in d au (B o d en se e) 20 M ü n ch en In g o ls ta d t, M ü n ch en , R o se n h ei m , L K A lt ö tt in g , L K B er ch te sg ad en er L an d , L K B ad T ö lz -W o lf ra ts h au se n, LK D ac h au , L K E b er sb er g , L K E ic h st ät t, L K E rd in g , L K F re is in g , L K F ü rs te n fe ld b ru ck , L K G ar m is ch - P ar te n ki rc h en , L K L an d sb er g (L ec h) , L K M ie sb ac h, L K M ü h ld o rf (I n n) , L K M ü n ch en , L K N eu b u rg - S ch ro b en h au se n, L K P fa ff en h o fe n (I lm ), LK R o se n h ei m , L K S ta rn b er g , L K T ra u n st ei n, L K W ei lh ei m - S ch o n g au , L an d sh u t, L K K el h ei m , L K L an d sh u t, L K R o tt al -I n n, L K D in g o lfi n g -L an d au , A u g sb u rg , K au fb eu re n, K em p te n (A llg äu ), M em m in g en , L K A ic h ac h -F ri ed b er g , L K A u g sb u rg , L K D ill in g en (D o n au ), LK G ü n zb u rg , L K O st al lg äu , L K U n te ra llg äu , L K D o n au -R ie s, L K O b er al lg äu 21 P as sa u P as sa u, L K D eg g en d o rf , L K F re yu n g -G ra fe n au , L K P as sa u, L K R eg en 22 R eg en sb u rg S tr au b in g , L K S tr au b in g -B o g en , R eg en sb u rg , W ei d en (O b er p fa lz ), LK C h am , L K N eu st ad t ( W al d n aa b) , LK R eg en sb u rg , L K S ch w an d o rf 1 LK s ta nd s fo r La nd kr ei se , w hi ch a re d is tr ic ts c o m b in ed fr o m d iff er en t ( m o st ly ) s m al le r m un ic ip al iti es . • Patrizio Vanella, Timon Hellwagner, Philipp Deschermeier482 Ta b . A 3: C o n tin ua tio n N u m b er N am e D is tr ic ts 1 23 N ü rn b er g A m b er g , L K A m b er g -S u lz b ac h, L K N eu m ar kt (O b er p fa lz ), LK T ir sc h en re u th , B am b er g , B ay re u th , C o b u rg , LK B am b er g , L K B ay re u th , L K C o b u rg , L K F o rc h h ei m , L K K ro n ac h, L K K u lm b ac h, L K L ic h te n fe ls , LK W u n si ed el (F ic h te lg eb ir g e) , A n sb ac h, E rl an g en , F ü rt h, N ü rn b er g , S ch w ab ac h, L K A n sb ac h, LK E rl an g en -H ö ch st ad t, L K F ü rt h, L K N ü rn b er g er L an d , L K N eu st ad t ( A is ch -B ad W in sh ei m ), LK R o th , LK W ei ß en b u rg -G u n ze n h au se n, L K S o n n eb er g 24 S ch w ei n fu rt /W ü rz b u rg S ch w ei n fu rt , W ü rz b u rg , L K B ad K is si n g en , L K R h ö n -G ra b fe ld , L K H aß b er g e, L K K it zi n g en , LK M ai n -S p es sa rt , L K S ch w ei n fu rt , L K W ü rz b u rg 25 S aa rb rü ck en LK K ai se rl au te rn , P ir m as en s, Z w ei b rü ck en , K ai se rl au te rn , L K K u se l, LK S ü d w es tp fa lz , LK R eg io n al ve rb an d S aa rb rü ck en , L K M er zi g -W ad er n, L K N eu n ki rc h en , L K S aa rl o u is , L K S aa rp fa lz -K re is , LK S t. W en d el 26 B er lin B er lin , B ra n d en b u rg (H av el ), Fr an kf u rt (O d er ), P o ts d am , L K B ar n im , L K D ah m e- S p re ew al d , L K H av el la n d , LK M är ki sc h -O d er la n d , L K O b er h av el , L K O d er -S p re e, L K O st p ri g n it z- R u p p in , L K P o ts d am -M it te lm ar k, LK P ri g n it z, L K T el to w -F lä m in g , L K U ck er m ar k 27 R o st o ck R o st o ck , L K R o st o ck 28 N eu b ra n d en b u rg LK M ec kl en b u rg is ch e S ee n p la tt e 29 G re if sw al d /S tr al su n d LK V o rp o m m er n -R ü g en , L K V o rp o m m er n -G re if sw al d 30 C h em n it z LK H o f, H o f, C h em n it z, L K E rz g eb ir g sk re is , L K M it te ls ac h se n, L K V o g tl an d kr ei s, L K Z w ic ka u 31 D re sd en C o tt b u s, L K E lb e- E ls te r, LK O b er sp re ew al d -L au si tz , L K S p re e- N ei ß e, D re sd en , L K B au tz en , L K G ö rl it z, LK M ei ß en , L K S äc h si sc h e S ch w ei z- O st er zg eb ir g e 32 Le ip zi g Le ip zi g , L K L ei p zi g , L K N o rd sa ch se n, D es sa u -R o ß la u, H al le (S aa le ), LK A n h al t- B it te rf el d , LK B u rg en la n d kr ei s, L K M an sf el d -S ü d h ar z, L K S aa le kr ei s, L K W it te n b er g , L K A lt en b u rg er L an d 33 M ag d eb u rg M ag d eb u rg , L K A lt m ar kk re is S al zw ed el , L K B ö rd e, L K H ar z, L K J er ic h o w er L an d , L K S al zl an d kr ei s, LK S te n d al 34 E rf u rt E rf u rt , G er a, J en a, S u h l, W ei m ar , E is en ac h, L K W ar tb u rg kr ei s, L K U n st ru t- H ai n ic h -K re is , LK K y ff h äu se rk re is , L K S ch m al ka ld en -M ei n in g en , L K G o th a, L K S ö m m er d a, L K H ild b u rg h au se n, LK Il m -K re is , L K W ei m ar er L an d , L K S aa lf el d -R u d o ls ta d t, L K S aa le -H o lz la n d -K re is , L K S aa le -O rl a- K re is , LK G re iz 1 LK s ta nd s fo r La nd kr ei se , w hi ch a re d is tr ic ts c o m b in ed fr o m d iff er en t ( m o st ly ) s m al le r m un ic ip al iti es . Past and Future Trends in Refugee Migration on the Regional Level in Germany • 483 Fe d er al S ta te Y ea r 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 20 20 3. 41 3 2. 56 1 9. 42 5 0. 94 5 21 .0 61 7. 45 1 4. 83 2 13 .0 4 4 15 .6 21 1. 19 9 5. 14 3 3. 01 4 1. 97 2 4. 96 6 2. 72 4 2. 62 8 20 21 3. 42 1 2. 56 5 9. 4 40 0. 92 7 21 .0 35 7. 45 8 4. 83 9 13 .0 75 15 .6 77 1. 19 5 5. 14 9 3. 01 0 1. 96 0 4. 94 2 2. 69 7 2. 61 0 20 22 3. 42 9 2. 56 8 9. 45 6 0. 90 7 21 .0 09 7. 46 5 4. 84 7 13 .1 06 15 .7 34 1. 19 2 5. 15 5 3. 00 6 1. 94 8 4. 91 7 2. 67 0 2. 59 1 20 23 3. 43 7 2. 57 1 9. 47 1 0. 88 7 20 .9 84 7. 47 2 4. 85 4 13 .1 38 15 .7 9 1. 18 8 5. 16 1 3. 00 2 1. 93 6 4. 89 3 2. 64 4 2. 57 2 20 24 3. 4 45 2. 57 5 9. 48 6 0. 86 5 20 .9 58 7. 47 9 4. 86 2 13 .1 69 15 .8 47 1. 18 5 5. 16 7 2. 99 8 1. 92 5 4. 86 8 2. 61 8 2. 55 4 20 25 3. 45 3 2. 57 8 9. 50 1 0. 84 3 20 .9 32 7. 48 7 4. 86 9 13 .2 01 15 .9 04 1. 18 1 5. 17 3 2. 99 3 1. 91 3 4. 84 4 2. 59 2 2. 53 6 20 26 3. 46 1 2. 58 1 9. 51 7 0. 82 0 20 .9 07 7. 49 4 4. 87 6 13 .2 32 15 .9 61 1. 17 8 5. 17 9 2. 98 9 1. 90 1 4. 82 0 2. 56 6 2. 51 8 20 27 3. 46 8 2. 58 5 9. 53 2 0. 79 5 20 .8 81 7. 50 1 4. 88 4 13 .2 64 16 .0 19 1. 17 5 5. 18 5 2. 98 5 1. 89 0 4. 79 6 2. 54 0 2. 50 0 20 28 3. 47 6 2. 58 8 9. 54 7 0. 77 0 20 .8 55 7. 50 8 4. 89 1 13 .2 95 16 .0 76 1. 17 1 5. 19 1 2. 98 1 1. 87 8 4. 77 3 2. 51 5 2. 48 2 20 29 3. 48 4 2. 59 2 9. 56 3 0. 74 4 20 .8 30 7. 51 6 4. 89 9 13 .3 27 16 .1 34 1. 16 8 5. 19 7 2. 97 7 1. 86 7 4. 74 9 2. 49 0 2. 46 4 20 30 3. 49 3 2. 59 5 9. 57 8 0. 71 6 20 .8 04 7. 52 3 4. 90 6 13 .3 59 16 .1 92 1. 16 5 5. 20 3 2. 97 3 1. 85 6 4. 72 5 2. 46 6 2. 4 46 Ta b . A 4: P re d ic te d d is tr ib u tio n o f r ef ug ee s am o ng t he fe d er al s ta te s [in % ] S o ur ce : A ut ho rs ’ c o m p ut at io n an d d es ig n. T he n um b er in g o f t he fe d er al s ta te s is il lu st ra te d in F ig ur e 8. • Patrizio Vanella, Timon Hellwagner, Philipp Deschermeier484 D ep en d en t v ar ia b le : fi rs t p ri n ci p al c o m p o n en ts o f a n n u al g en d er -n at io n al it y em p lo ym en t c h an g es (1 ) (2 ) (3 ) (4 ) (5 ) (6 ) A fr ic an A si an o th er s fe m al e m al e fe m al e m al e fe m al e m al e R ef u g ee in fl o w -0 .0 41 ** * -0 .0 87 ** * -0 .0 17 ** * -0 .0 33 ** * -0 .0 86 ** * -0 .0 59 ** * (0 .0 02 ) (0 .0 05 ) (0 .0 01 ) (0 .0 02 ) (0 .0 08 ) (0 .0 06 ) G D P, g ro w th r at e -1 ,6 49 .5 95 -1 4, 32 5. 73 0* ** -2 ,8 64 .1 82 -5 ,9 28 .7 81 -2 ,2 46 .4 71 -4 ,2 98 .7 11 * (1 ,0 51 .8 05 ) (4 ,0 72 .8 35 ) (5 ,0 69 .2 15 ) (1 1, 46 4. 30 0) (1 ,4 51 .2 09 ) (2 ,2 70 .5 98 ) C o n st an t -1 67 .0 32 ** * 16 4. 21 4 -8 43 .2 35 ** * -2 80 .8 59 -4 30 .7 78 ** * -4 20 .6 82 ** * (3 8. 18 3) (1 47 .3 32 ) (1 77 .6 50 ) (3 99 .8 94 ) (4 8. 84 0) (7 6. 4 41 ) O b se rv at io n s 12 12 12 12 12 12 R 2 0. 97 0. 98 0. 94 0. 98 0. 93 0. 92 A d ju st ed R 2 0. 97 0. 97 0. 93 0. 97 0. 92 0. 90 R es id u al S td . E rr o r (d f = 9 ) 92 .9 35 9. 5 4 47 .4 1, 01 1. 6 12 8. 1 20 0. 4 F S ta ti st ic (d f = 2 ; 9) 15 3. 5* ** 18 1. 7* ** 70 .2 ** * 20 8. 2* ** 63 .4 ** * 49 .2 ** * Ta b . A 5: R eg re ss io n o u tp u t * p < 0 .1 , * * p < 0 .0 5, * ** p < 0 .0 1 S o ur ce : A ut ho r's c o m p ut at io n an d th e so ur ce s o ut lin ed in th e m ai n te xt . Published by Federal Institute for Population Research (BiB) D-65180 Wiesbaden / Germany Managing Publisher Dr. Nikola Sander 2022 Editor Prof. Frans Willekens Managing Editor Dr. Ralina Panova Dr. Katrin Schiefer Editorial Assistant Beatriz Feiler-Fuchs Wiebke Hamann Layout Beatriz Feiler-Fuchs E-mail: cpos@bib.bund.de Scientifi c Advisory Board Karsten Hank (Cologne) Ridhi Kashyap (Oxford) Michaela Kreyenfeld (Berlin) Natalie Nitsche (Rostock) Zsolt Spéder (Budapest) Alyson van Raalte (Rostock) Rainer Wehrhahn (Kiel) Comparative Population Studies www.comparativepopulationstudies.de ISSN: 1869-8980 (Print) – 1869-8999 (Internet) Board of Reviewers Bruno Arpino (Barcelona) Kieron Barclay (Rostock) Laura Bernardi (Lausanne) Gabriele Doblhammer (Rostock) Anette Eva Fasang (Berlin) Michael Feldhaus (Oldenburg) Tomas Frejka (Sanibel) Alexia Fürnkranz-Prskawetz (Vienna) Birgit Glorius (Chemnitz) Fanny Janssen (Groningen) Frank Kalter (Mannheim) Stefanie Kley (Hamburg) Bernhard Köppen (Koblenz) Anne-Kristin Kuhnt (Duisburg) Hill Kulu (St Andrews) Nadja Milewski (Wiesbaden) Roland Rau (Rostock) Thorsten Schneider (Leipzig) Tomas Sobotka (Vienna) Jeroen J. A. Spijker (Barcelona) Heike Trappe (Rostock) Helga de Valk (The Hague) Sergi Vidal (Barcelona) Michael Wagner (Cologne) http://www.comparativepopulationstudies.de mailto:cpos@bib.bund.de