Local Labour Demand and Immigrant Employment

Journal Pre-proof

Local Labour Demand and Immigrant Employment Luz Azlor , Anna Piil Damm , Marie Louise Schultz-Nielsen PII: DOI: Reference:

S0927-5371(20)30014-2 https://doi.org/10.1016/j.labeco.2020.101808 LABECO 101808

To appear in:

Labour Economics

Received date: Revised date: Accepted date:

14 September 2018 30 January 2020 31 January 2020

Please cite this article as: Luz Azlor , Anna Piil Damm , Marie Louise Schultz-Nielsen , Local Labour Demand and Immigrant Employment, Labour Economics (2020), doi: https://doi.org/10.1016/j.labeco.2020.101808

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2020 Published by Elsevier B.V.

Highlights

 This paper investigates the effect of local labour demand on employment of immigrant workers. We take into account self-selection into locations by estimating the effects for refugees who were subject to the Danish Spatial Dispersal Policy from 1999-2010 using full population Danish administrative registers that contain information on admission class of immigrants. We identify refugee status without any measurement error. Our findings show that assignment to a municipality with a one percentage point higher employment rate increases the employment probability of refugees by 0.5-0.6 percentage points (elasticities 1.5-2.3) two to four years after asylum in Denmark, while assignment to a municipality with a one percentage point higher unemployment rate decreases the individual employment probability of refugees by 0.9-1.7 percentage points (elasticities -3.8--3.2) two to four years after asylum.We also find significant effects of alternative measures of local labour demand on employment chances of refugees. Our results provide quasi-experimental evidence that immigrant employment is more sensitive to local labour demand conditions than natives.

1

Local Labour Demand and Immigrant Employment1 Luz Azlor,2 Anna Piil Damm3 and Marie Louise Schultz-Nielsen4 January 30th 2020 Abstract: This study investigates the effect of local labour demand on employment of immigrant workers. We address the challenge of location sorting by estimating the effects of initial local labour demand for refugees who were subject to the Danish Spatial Dispersal Policy from 1999-2010. After location assignment, refugees participate in a 3-year introduction programme; eligibility to means-tested welfare benefits during programme participation is conditional on residing in the assigned municipality. We use full population Danish administrative registers that contain information on admission class of immigrants and identify refugee status without any measurement error. Our findings show that four years after assignment, 83% of refugees still live in the assigned municipality. Moreover, assignment to a municipality with a one percentage point higher employment rate increases the employment probability of refugees by 0.5-0.6 percentage points (elasticities 1.1-1.8) two to four years after arrival in Denmark. We also find consistent significant effects of alternative measures of local labour demand on employment chances of refugees. Our results provide quasiexperimental evidence that immigrant employment is sensitive to labour market conditions in the initial location and highlight the importance of carefully designing refugee allocation policies. Keywords: Immigrants, Refugees, Asylum Seekers, Settlement Policies, Employment, Natural Experiment. JEL codes: J23, J61, J68, J71

I. INTRODUCTION Increasing rates of immigrants over the past decades in Western countries have spurred debates about immigration and integration policies, questioning whether the host economies can successfully integrate immigrants into the labour market (Bauer, Lofstrom and Zimmermann 2000; Dustmann, Vasiljeva and Damm 2019). Two recent major events have sparked the debate in Europe. The 2004 and 2007 enlargements of the common European labour market,

1

This paper acknowledges the support from and access to Statistics Denmark provided by the ROCKWOOL Foundation Research Unit. We acknowledge financial support from the ROCKWOOL Foundation Research Unit, TrygFonden and the Department of Economics and Business Economics, Aarhus University. We thank the editor and two anonymous reviewers for helpful comments and Peter Fredriksson for his comments on an early draft. We thank Bente Herbst Bendiksen and Janne Lindblad at the Danish Immigration Service for sharing their internal administrative statistics and knowledge about the Danish Spatial Dispersal Policy 1999-2016 with us. Finally, we thank Mie Hjortskov Andersen, Drilon Helshani, Villiam Vellev and Camilla Hedegaard Hansen for research assistance. The authors declare that they have no relevant or material financial interests that relate to the research described in this paper. 2 Department of Economics and Business, Aarhus University, Fuglesangs Allé 4, DK-8210 Aarhus V. Email: [email protected]. Current work place: EDI Global, Prospect House, High Wycombe, HP13 6LA, UK. Email: [email protected]. 3 Department of Economics and Business, Aarhus University, Fuglesangs Allé 4, DK-8210 Aarhus V. Email: [email protected]. 4 ROCKWOOL Foundation Research Unit, Ny Kongensgade 6, 1472 København K. Email: [email protected].

2

which triggered a massive inflow of labour migrants from Eastern Europe to the old EU countries, and the sudden increase in the influx of refugees, notably from Syria, to EU countries, which culminated in the autumn 2015. In a world with large cross-country productivity differences, there is a potential for substantial economic gains from immigration, as open borders allow labour to flow towards its best use (Kennan 2013; Bratsberg, Raaum and Røed 2017). Moreover, immigration of labour may alleviate the demographic and fiscal challenges that European countries with ageing populations face (Storesletten 2000; Bratsberg et al. 2017). Although employment and earnings of immigrants increase with years spent in the host country (Chiswick 1978; Borjas 1985; LaLonde and Topel 1992; Dustmann 1993; Borjas 1995; Lubotsky 2007; Algan, Dustmann, Glitz and Manning 2010; Sarvimäki 2011, Dustmann and Görlach 2015), studies have documented substantial employment and earnings disparities between immigrants and natives, and in particular for non-labour migrants, the immigrant-native employment gap remains large even 7-10 years after immigration (Edin, Fredriksson and Åslund 2003; Damm 2009a; Damm and Rosholm 2010; Bratsberg et al. 2017; Schultz-Nielsen 2017; Ruiz and Vargas-Silva 2018; Fasani, Frattini and Minale 2018; Schultz-Nielsen 2019).5 6 A substantial part of the literature on immigrant employment has investigated the importance of supply-side factors such as admission class, skills acquired in the host country, potential work experience and language ability (Chiswick 1978; Borjas 1985; Borjas 1995; Husted, Nielsen, Rosholm and Smith 2001; Cortes 2004; Algan, Dustmann, Glitz and Vogel 2010; Schultz-Nielsen 2016). The aim of our study is to estimate how sensitive immigrant employment is to local labour market conditions. Immigrant employment may in fact be more sensitive to changes in local labour market conditions than native employment. This may be the case for at least four reasons. First, immigrants are overrepresented in low skilled jobs (Smith et al. 2003; Edin, Fredriksson and Åslund 2004) which fuels instability, in part due to skill- or routine-biased technological change (Katz and Murphy 1992; Berman, Bound and Machin 1998; Card and DiNardo 2002; Moore and Ranjan 2005; Goos, Manning and Salomons 2014) and task offshoring (Grossman and Rossi-Hansberg 2008; Goos et al. 2014).7 Second, employers may discriminate immigrant applicants.8 Third, in many countries firms use the last-in-first-out (LIFO) principle in downsizing and immigrants are likely to be overrepresented in the group of recent hires (Bratsberg et al. 2017; Åslund et al. 2017). Fourth, job-referral networks of recent immigrant cohorts are ethnically stratified.9 The question of how sensitive immigrant employment is to local labour demand conditions is important for several reasons. First, such knowledge will give us an understanding of the extent to which economic growth alone can increase the employment rate of immigrants. Second, such knowledge can be used for optimal design of public 5

The non-Western immigrant-native employment gaps are particularly large in the Nordic countries, partly due to the high labour force participation rates of natives (including that of women) on which the Nordic welfare models rely (Bratsberg et al. 2014; Bratsberg et al. 2017; Schultz-Nielsen 2017; Åslund, Forslund and Liljeberg 2017; Sarvimäki 2017). 6 See Becker and Ferrara (2019) for a recent review of consequences of forced migration. 7 In Denmark, as in many other countries, the largest share of non-Western immigrants work in the service industry (around 32%), which has not been affected by routine-biased technological changes and task offshoring to the same extent as the manufacturing industry. Across industries in Denmark over the 2008-2016-period, the manufacturing industry has experienced the largest reduction in the share of workers: 2.2 percentage points, which given its employment share of 13.8% in 2008 corresponds to a 15.8% reduction. With an employment share of 14.3% in manufacturing, non-Western immigrants were slightly overrepresented relative to natives in 2008, but with a share of only 10.4% in manufacturing in 2016, non-Western immigrants are underrepresented in manufacturing. (Authors’ own calculations from Danish public employment statistics across industries and immigrant status, http://www.statistikbanken.dk/RAS311). 8 For empirical evidence from correspondence studies of discrimination by ethnic origin, see Riach and Rich (1991), Esmail and Everington (1997), Bertrand and Mullainathan (2004), Carlsson and Rooth (2007). 9 For causal evidence for the U.S., see Munshi (2003) and Beaman (2012). For causal evidence for Scandinavia, see Edin, Fredriksson and Åslund (2003) and Damm (2009, 2014).

3

employment policies. For instance, during economic recession it may be optimal to increase resources for employment programs for immigrants or low-skilled workers to stimulate local demand for their skills and spend additional resources on training and skill-upgrading programs. Third, such knowledge can be used for optimal design of settlement policies on newly recognized refugees and asylum seekers. In particular, current settlement policies on refugees employed in a number of European countries can be reformed in order to increase the speed of labour market integration of refugees. Our empirical approach addresses the challenge that immigrants self-select into locations (see e.g. Edin, Fredriksson and Åslund 2003; Damm 2009a; Damm 2014). If people with favourable characteristics sort themselves into locations with favourable labour market conditions, point estimates from observational data studies will overestimate the true effect in absolute terms. To get exogenous variation in local labour market conditions, we exploit the current Danish Spatial Dispersal Policy on Refugees in place since 1999. The allocation is based on a quota-system that distributes newly arrived refugees to municipalities with lower shares of foreigners. Our study is the first to exploit this policy for identification of causal effects of local characteristics on integration of immigrants into the host country society. In particular, we exploit the policy to identify the effects of the initial local labour demand in the assigned location on immigrant employment. We provide consistent estimates of the effects of initial local labour market conditions in the assigned location by identifying refugee status without any measurement error and by using the subsample of refugees who received asylum after a substantial share of municipalities had reached full capacity in a given year. Our study provides quasi-experimental estimates on the sensitivity of immigrant employment to the local employment rate. We find that a one percentage point higher employment rate in the assigned location increases the employment probability of refugees by 0.5-0.6 percentage points (elasticities 1.1-1.8) two to four years after arrival in Denmark. We also estimate the effect of the initial local unemployment rate on the employment probability of refugees. We find that assignment to a municipality with a one percentage point higher unemployment rate decreases the individual employment probability of refugees by 0.9-1.7 percentage points (elasticities -0.16-(-0.19)) two to four years after asylum. Our study is related to three existing impact evaluations that exploit spatial dispersal policies on refugees in Scandinavia to identify the causal effect of local labour market conditions on immigrant labour market outcomes: Åslund and Rooth (2007) for Sweden, Damm and Rosholm (2010) for Denmark and Godøy (2017) for Norway. While our study provides quasi-experimental estimates of the effects of the initial employment rate in the assigned location as well as for a range of alternative measures of local labour market conditions on immigrant employment, Åslund and Rooth (2007) estimates the effects of the initial local unemployment rate on both immigrant earnings and employment. Godøy (2017), instead, estimates the effects of the initial regional employment rate of non-OECD immigrants on immigrant earnings, whereas Damm and Rosholm (2010) exploits the first Danish Spatial Dispersal Policy on refugees in place from 1986-1998 to provide quasi-experimental evidence on local determinants on the hazard rate into first job. An important strength of our paper relative to Åslund and Rooth (2007) and Damm and Rosholm (2010) is that we identify refugee status without any measurement error using Danish administrative registers from 1999. Our study hereby addresses an important concern in previous studies that the estimated effects of local labour market conditions are biased due to measurement error stemming from potential use of a contaminated sample. To our knowledge, only one existing study (Åslund and Rooth, 2007) provides empirical evidence on the direction of the bias of estimates of the effect of local labour demand conditions on immigrant labour market outcomes in observational data studies. Comparison of their point estimates with point estimates using an observational data approach suggests that observational data studies tend to overestimate the true effect in absolute terms, because 4

immigrants with favourable characteristics tend to self-select into locations with favourable labour market conditions. We investigate whether this result generalizes to another Scandinavian country and find that it does. Damm and Rosholm (2010) find that being assigned to a region of high unemployment delays the employment of refugees. Using the second Danish Spatial Dispersal Policy in place since 1999, we test whether their finding generalizes to a later time period and confirm that it does. Our study also contributes to the economic literature by showing how researchers can make better use of micro data for refugees subject to a Spatial Dispersal Policy due to municipal quotas, including the ‘Whole of Sweden’ Strategy used in e.g. Edin et al. (2003) and Åslund and Rooth (2007) and the Danish Spatial Dispersal Policy in place since 1999. In future research, instead of using the full population of refugees subject to the Spatial Dispersal Policy, consistency of estimated effects of location characteristics can be increased by extracting the subsample of refugees who received asylum after a substantial share of municipalities had reached full capacity in a given year. The economic literature on the effects of local labour demand conditions on immigrant labour market outcomes is rather limited. Observational data studies suggest that ethnic minority groups are more sensitive to changes in labour market conditions than the majority population (see Hoynes 2000 for U.S. evidence.10 For Norwegian evidence, see Barth, Bratsberg and Raaum 2004; Bratsberg, Raaum and Røed 2010; Bratsberg et al. 2017).11 Our estimates do not suffer from location sorting at the time of arrival to the host-country. Therefore, our point estimates of the effect of the local unemployment rate on the employment probability of refugees are much smaller in absolute terms than in observational data studies (e.g. Bratsberg et al. 2017). In fact, (in absolute terms) our point estimates are also lower than the point estimates 2-4 years after arrival estimated by Åslund and Rooth (2007).12 The structure of our paper is as follows. Section II describes the institutional background. In Section III we provide our methodological considerations and set up our empirical model. Then follows a description of our data in Section IV, and a presentation of our empirical results in Section V. In Section VI, we discuss our results and provide our concluding remarks.

II. INSTITUTIONAL BACKGROUND People seeking asylum in Denmark need to register at the Police upon arrival and be interviewed by the Danish Immigration Service (DIS). Depending on the capacity of DIS at the time, the time between the registration day and the first interviewed may vary. Until DIS has a verdict on the asylum application, asylum seekers are assigned to asylum centres. On average, Hvidtfeldt and Schultz-Nielsen (2018) calculate that in the period of 1997-2011, the 10

Using a relatively weak identifying assumption (relying only on within-county variation), Hoynes (2000) estimates the effects of different measures of local labour demand on transitions off welfare and transitions back to welfare using administrative data for California. She finds that Hispanics, Blacks, residents of urban areas and unemployed parent recipients of welfare are more sensitive to changes in labour market conditions while whites and teen parents are less sensitive. 11 Examining labour market integration across admission classes of immigrants in Norway, Bratsberg et al. (2017) find that a one percentage point increase in the local unemployment rate is associated with a similar drop in the employment rate of male natives, but with a substantially higher drop in the employment rate of immigrants, except for West-European immigrants. For refugees, a one percentage point increase in the local unemployment rate decreases their employment probability by 5 percentage points, and the estimates are similar for male and female refugees. 12 This difference may arise from differences in our sampling strategies, most importantly the strength of our study due to perfect identification of refugee status in our data and use of a subsample of refugees who got asylum after a substantial share of municipalities had reached their annual capacity.

5

application process took 15 months. During this waiting time, asylum seekers are provided basic needs, such as housing, food and clothing, but they have limited or no access to the Danish labour market (Dustmann et al. 2017). As a matter of fact, it was not until mid-2013 that asylum seekers were allowed to be employed, and even after the reform, bureaucratic obstacles still exist, making it possible only for a handful of asylum seekers to be working during the waiting period (Danish Broadcasting Corporation 2016).13 Moreover, because most asylum centres are located in remote areas, the residents have limited contact with the Danish society. Hence, when we refer to initial local labour market in this study it is the municipality of assignment to which asylum seekers are allocated if they obtain a residence permit. Denmark has had spatial dispersal policies for refugees who had their applications approved since 1986 (Damm 2005).14 The purpose has been to disperse refugees equally across municipalities to ensure an even distribution of the integration task across the country and to avoid settlement of newly arrived refugees in areas in which the concentration of foreign nationals is already high, which could potentially hinder refugees’ introduction to the Danish language or society in general. Until 1998, the Danish Refugee Council (DRC) organized the spatial dispersion of the refugees and was in charge of the 18 months long introduction program that included courses in Danish language, culture and job training. Despite the goal of the placement being to distribute refugees equally in proportion to the population size, to promote ethnic networks, refugees were spatially dispersed in clusters with fellow countrymen. Although the refugees were encouraged to stay in the municipality of assignment, eligibility for means-tested social benefits was not conditional on the refugees staying there (Ibid 2005). With the Danish Parliament’s enactment of the ‘Integration Law’ from the 1st of Jan. 1999, the introduction programme (consisting of courses on the host-country society and language) was extended from 18 months to 3 years and the responsibility for both this programme and the spatial dispersal policy reception was handed over to the municipalities hosting the refugees. The new legislation further strengthened the effort to allocate refugees across all municipalities in the country and made eligibility to means-tested welfare benefits during the 3-year introduction programme conditional on residing in the assigned municipality. A study covering the years 1997-2005 finds that the reform succeeded in distributing refugees more evenly across municipalities relative to the local population size (and away from larger cities) and reduced refugees’ tendency to move out of the municipality of assignment (Nielsen and Jensen 2006). Over the period 1997-2005, the average yearly out-migration rate from the municipality of assignment fell from 17% before the reform to 7% after, and post-reform movers relocated after more years since asylum than pre-reform movers. The geographical mobility pattern of refugees arrived from 1999-2010 is shown in Figure 1. The dotted line illustrates the share of refugees that still live in the municipality of assignment bi-annually since asylum. The solid line shows the percentage of refugees that out-migrate from the municipality of assignment bi-annually since asylum. During the first three years after asylum, the bi-annual out-migration rate from the municipality of assignment is very low (0.4-2.5%) and by the end of the third year, 89% of the refugees are still living in the municipality of assignment. The low relocation rate is related to the requirement of staying in the municipality of assignment in order to be eligible for 13

https://www.dr.dk/ligetil/kun-31-asylansoegere-kom-i-job-sidste-aar. Thus, a refugee’s access to the Danish labour market is delayed until the refugee obtains a residence permit. Hvidtfeldt et al. (2018) control for this delay and find no significant employment effect of waiting time for refugees in Denmark for the period 1997-2014. 14 People that apply for asylum in Denmark after travelling independently are regarded as spontaneous asylum seekers and a fraction of them have their applications approved and are granted asylum. Additionally, from 1989 until 2016, the UNHCR resettled close to 500 refugees annually in Denmark, which Danish authorities call “quota refugees”. Henceforth, we refer to any such recognized refugees, whether spontaneous or quota, as “refugees”. Around 34,000 refugees were granted asylum in Denmark from 1997 to 2011 among whom 21% where quota refugees (Hvidtfeldt and Schultz-Nielsen 2018).

6

means-tested benefits during the 3-year introduction programme period. As soon as this period ends, the relocation rate increases to 8.3 % in first half of the fourth year since asylum but declines thereafter. Hence, the postponed desire of moving that some refugees seem to have is fulfilled rather quickly. By the end of the fourth year since asylum, 76 % of the refugees are still living in the municipality of assignment. Figure 1 documents that the spatial dispersal policy in place since 1999 influences refugees’ settlement for a long period. [Figure 1. Stability of settlement for refugees arriving 1999-2010] The legal basis for the spatial dispersal policy is stipulated in the third chapter of the ‘Integration Law’. It specifies that The Danish Immigration Service (DIS) each year shall make a forecast (‘Landstallet’ in Danish) of the overall number of refugees that are expected to obtain refugee status in the following calendar year. Based on this forecast, allocation of refugees to municipalities is settled in agreement between these local authorities.15 The allocation is based on a quota-system that distributes newly arrived refugees to municipalities with lower shares of foreigners. The annual quota reflects the municipality’s share of the total population adjusted for the share of foreigners already living in the municipality.16 Appendix A gives an accurate description of the formula for calculation of the annual municipal quota, given the forecast of the overall number of recognized refugees in the calendar year. Overall, this calculation method has remained unchanged from 1999 to 2016.17 Using the formula to predict the annual municipal refugee quotas, we find a correlation between the predicted and actual annual refugee quota across municipalities as high as 0.96 over the period from 1999 to 2016; the correlation between the actual and predicted refugee quota for each municipality for 2010 is illustrated in Figure A1 in the Appendix. Such a high correlation provides strong evidence that in spite of the possibility for municipalities to negotiate the annual refugee quota, the actual annual municipal refugee quotas are close to the quotas that DIS would set in the absence of municipal agreement. It is the responsibility of DIS to refer each refugee to a municipality that has not yet met its yearly quota. During the asylum process, a caseworker from DIS meets the asylum seeker, first and foremost to confirm the identity of the asylum seeker and asks other questions related to the asylum case. During the meeting, the asylum seeker can also express wishes to be settled in a particular location to pursue education or health treatment or to live close to relatives, if granted residence (DIS-interview18).19 While DIS tries to satisfy those preferences, only under extraordinary circumstances will a refugee be referred to a municipality with a full quota. Close family already living in Denmark is primarily considered and spouses and children are always settled in the same municipality as the first arrived family member. Other conditions that can be taken into consideration are nationality and thereby the refugee’s possibility of creating a network with countrymen, educational qualifications and critical need of (medical) treatment (“Order/BEK # 630 of 25/08/1998”). Municipalities, on their side, may also have wishes regarding the educational background of the refugees they will receive. However, educational qualifications from the refugees’

15

If the local authorities do not reach an agreement, DIS determines the allocation based on the calculated quotas. DIS may adjust the quotas (proportionally) if the actual number of refugees differs substantially from the expected number. Throughout our analysis, we use for any given year the announced municipal quota that was valid on the 31st of December of the year before. 17 From 1999 to July 2016 the share of foreigners includes foreign nationals, except those from Nordic countries, EU/EEA. After July 2016 the definition of foreigners used for the quota-calculation is slightly changed and placement should include employment considerations (“Order/BEK # 980 of 28/06/2016”). 18 th The authors conducted an interview that addressed the administration of the Danish Spatial Dispersal Policy in January 18 , 2017 with Bente Herbst Bendiksen and Janne Lindblad at the Danish Immigration Service (DIS). 19 st This feature of the Danish Spatial Dispersal Policy on Refugees, in place since January 1 1999, was not part of the first Danish Spatial Dispersal Policy of Refugees in place from 1986-1998. Under the first policy, placement officers did not interview newly recognized asylum seekers and they assigned refugees to locations with little or no regard for location wishes (see Damm 2014; Damm and Dustmann 2014). 16

7

home countries are typically not easily transferred to the Danish labour market and the municipalities’ desire for special educational groups are, in any case, modest (DIS-interview). The specific date in which the household head has been assigned to the municipality is key for our identification. The date of municipality assignment is the date of residence permit (asylum) recorded in the residence permit register, provided by DIS. We also refer to this date as the date of arrival.20 With the exception of UN quota refugees (of which Denmark invited 500 annually until 2016), applicants for asylum apply after arrival to Denmark and live in a refugee reception centre until their asylum application has been processed. Figure 2 shows the accumulated number and share of municipalities that have reached their annual quota of refugees by a given calendar month for each year in our sample period, i.e. the cumulative distribution function of municipalities that have reached their full capacity. 21 The figure shows the following: First, that due to an unexpected high influx of asylum seekers in 1998, in 1999, the first 10 municipalities had reached full capacity in February, while in most calendar years, the first 10 municipalities had reached full capacity around June. Second, that at least 11 municipalities had reached their annual refugee quota by the fall of the year. Therefore, our preferred estimation sample includes observations from all calendar years. Third, in around half of the calendar years, a large fraction of municipalities met their annual refugee quota (1999, 2006, 2007, 2008, 2009). [Figure 2. Include around here] Hence, the yearly assigned municipality quotas are the core element in the Danish Spatial Dispersal Policy in place from 1999 until 2016. At the beginning of each calendar year, settlement of refugees is possible in all municipalities with a positive quota, but as the months pass and more refugees are granted residence, the municipal quotas become filled. Then, refugees’ potential preferences regarding the settlement can be more easily accommodated by DIS if the refugee obtains residence permit earlier in the calendar year rather than later in the same calendar year. If a refugee wishes to go to a municipality that has already fulfilled its annual quota, the individual will instead settle in one of the municipalities with vacant slots. Importantly, this aspect of the refugee settlement policy is a novel finding of our interview with DIS and has not been discussed in public. Besides, the date on which a refugee is assigned to a municipality can be considered outside the control of the refugee, given that the month in which asylum is claimed can be regarded as non-strategic and municipal assignment takes place shortly after receiving asylum and asylum seekers’ waiting period to obtain a Danish residence permit can last months (or even years). Thus, investigating in this context the effects of local labour market conditions for the subsample of refugees assigned to a municipality in the later months of the year resembles a field experiment.

III. METHODOLOGICAL CONSIDERATIONS AND EMPIRICAL MODEL III.A. Methodological considerations The main challenge in identification of the effects of local labour demand conditions on immigrant employment arises because immigrants may sort into locations in terms of individual characteristics, which are unobserved by 20

It takes on average 40 days from refugees are permitted residence until they are registered in the municipality population register in the period 2005-2010, for which the calculation has been made (Hvidtfeldt and Schultz-Nielsen, 2018). 21 We use the register information on the date of arrival of refugees to each municipality to calculate the number of assigned refugees in each municipality at any given date during a year and compare this number with the municipal refugee quota to construct the date at which each municipality reaches its annual quota.

8

researchers. Previous research has shown that this is indeed the case. Among refugees subject to a spatial dispersal program at the time of receiving asylum, individuals who subsequently moved into ethnic enclaves were negatively selected in terms of individual unobservable characteristics, for instance, English language proficiency (Edin et al. 2003; Damm 2009). Therefore, use of observational data for estimation of the effects of local labour demand conditions on immigrant employment will result in biased results due to omitted variables. Instead, we estimate the effects for the population of refugees who were assigned to housing across municipalities in Denmark upon being granted asylum after the first municipalities had reached the annual quota announced by the end of the previous calendar year. Local labour demand can be measured in different ways. The local unemployment rate is a common measure which reflects excess supply of labour at the minimum wage (according to the neoclassical theory of the firm). The local unemployment rate is also negatively correlated with labour market tightness defined as the number of job vacancies relative to the number of unemployed (according to job search theory). By definition, the unemployment rate is also negatively correlated with the labour force participation rate. In a situation with excess supply of labour, long-term unemployed workers may leave the work force as discouraged workers and re-enter the work force again when the local labour demand increases again. Thus, both the nominator and denominator in the unemployment rate would change, leaving the unemployment rate relatively unaffected by the number of discouraged workers. Moreover, longterm unemployed that are no longer entitled to unemployment insurance benefits have little financial incentive to stay in the work force. Unskilled as well as skilled workers with obsolete skills or, in case of immigrant workers, not easily transferable skills from their source countries, are likely to be overrepresented among those long-term unemployed and discouraged workers. In fact, the share of discouraged workers is non-negligible according to the OECD. Therefore, the local unemployment rate may not be an accurate measure of excess supply of workers in business cycles downturns. Alternatively, the local employment rate (defined as the number of employed relative to the population in the working ages) may be a better measure of local labour demand as only the nominator is affected by the number of discouraged workers.22 In other words, the employment rate varies with the number of discouraged workers. Our analyses concern local labour demand for a particular group of workers, namely immigrants from non-Western countries, who may bring skills that are not perfectly transferred to the host country’s labour market and may not be proficient in the host country’s language. For this reason, local demand for their labour may be better measured by the local unemployment rate and employment rate among non-Western immigrants, which we include as alternative measures of local labour demand in our analyses.23 The final measure of local labour demand we use is the net employment growth, which is equal to the difference between job creation and job destruction. We calculate it as the annual change in the number of employed individuals relative to the number of employed individuals last year.24 Similar to the employment rate, it varies with the number of discouraged workers. But in contrast to the unemployment and employment rates, it does not measure labour demand relative to the (potential) supply as measured by the (potential) size of the labour force. In the baseline regressions, we use the local employment rate to measure local labour demand due to our considerations described above and following Hoynes (2000), who concludes that employment-based measures of 22

Employment is an equilibrium outcome determined by the intersection between labour demand and labour supply. Since the number of discouraged workers affects the labour supply, the level of employment reflects the number of discouraged workers. 23 Western countries include the 28 EU-countries plus Andorra, Australia, Canada, Iceland, Liechtenstein, Monaco, New Zealand, Norway, San Marino, Switzerland and USA. Non-Western countries include all other countries (Statistics Denmark 2018). 24 This measure of local labour market conditions is inspired by Hoynes (2000).

9

Comment [COMP1]: This may be funding agency. Please check

local labour demand perform better than unemployment-based measures. We also use the four alternative measures of local labour demand to check the robustness of our baseline findings. Identification of the effects of local labour demand on immigrant employment requires control for correlated effects, that is, other characteristics of the local labour market that are correlated with both the local labour demand and individual employment, e.g. city size, job search networks and commuting costs. In the baseline specification, we control for the two municipality characteristics used by DIS to determine the annual municipal quota for the following year in combination with the expected number of refugees: the share of the Danish population living in the municipality and the non-Western immigrant share. In robustness checks, we include results after inclusion of additional controls. First, the co-national share as a measure of job search networks. Second, alternative measures of commuting costs to the centre of the commuting area (distance in kilometres, commuting time using public transportation and commuting time by car). Third, commuting area fixed effects.

III.B. Empirical model We now describe our empirical model for investigation of the importance of local labour market conditions for immigrants’ employment success. We avoid the standard identification challenge that arises because new immigrants self-select into locations by estimating the effects of local labour market conditions for refugees who get asylum after the first municipalities have reached their full capacity for that year and by conditioning on the characteristics of the household head which were observed by DIS at the time of assignment. Furthermore, to take account of possible selfselection of refugees into subsequent locations of residence, our baseline model explains refugee’s employment status in year t+s, ( ) by initial local labour market conditions in the municipality of assignment j*, ( ): (1)

(

)

( )

( )

(

)

where the subscripts denote i: individual, j: current municipality of residence, c: country of origin, t: year of asylum (municipal assignment), m: month of asylum, s: years since asylum. The parameter of interest is ̂ which provides an estimate of the intent-to-treat (ITT) effect of the local employment rate in the assigned municipality in year t on the individual’s employment in year t+s under the following assumptions: (i) refugees in our sample were randomly distributed across municipalities, conditional on the observed personal attributes, which were known by DIS and may therefore have affected the assignment to municipality type, (ii) there are no omitted correlated effects. Since the model excludes control for current labour demand conditions, satisfaction of the latter identifying assumption requires high correlation between the initial and current labour demand conditions. Evidence presented in Section IV lends empirical support to both identifying assumptions. In robustness checks, ̂ provides an estimate of the effect of each of the four alternative measures of initial local labour demand conditions: the local unemployment rate, the local unemployment rate among non-Western immigrants, the local employment rate among non-Western immigrants and the local employment growth in the assigned municipality in the year of assignment. represents personal characteristics at time t, ( ) represents characteristics of the municipality of assignment at time t (share of total population of the country and non-Western immigrant share in the local population), while is country of origin fixed effects, just like and are,

10

respectively, year and month of asylum fixed effects. ( ) is the error term. We estimate the model by OLS for each year since asylum s=2, 3, 4.25 Henceforth, we refer to the model in Equation 1 as the baseline model.26 27 We exploit between and within municipality variation in local labour market conditions across 95 out of 98 municipalities; the remaining 3 municipalities (including Copenhagen Municipality) had a zero annual quota of refugees throughout our sample period, 1999-2010. Since the treatment – municipal labour market conditions – varies by municipality, we cluster the standard errors by municipality of assignment. In order to investigate the direction of the bias of estimates from observational data approaches, we estimate a second model. It mirrors the observational data approach by describing the association between the immigrant’s current employment status and current local labour market conditions, controlling for initial personal characteristics and other current municipality characteristics in a given year since assignment: (2)

(

)

(

)

(

)

(

)

where

( ) is the error term. The remaining variables and indices in Eq. 2 are the same as in Eq. 1, except that represents other characteristics of the municipality of residence at time t+s (share of total population of the ( ) country and non-Western immigrant share in the local population). The parameter of interest, ̂, provides an observational data estimate of the effect of the local employment rate in the municipality of residence in year t+s, ( ) . ̂ will only provide a consistent estimate of the (average treatment) effect of the local labour market characteristics under the strong assumption of no self-selection of immigrants into locations in year t+s and no omitted correlated effects.

To mirror the observational data approach more closely, we estimate Eq. 2 using a balanced panel of refugees who are subject to the Danish Dispersal Policy that assigns them to a municipality upon arrival (at time t). ̂ estimated on such a sample will be biased both in case of initial self-selection into locations of early arrivals during the calendar year and subsequent self-selection into location. To isolate the bias due to the latter, we also estimate Eq. 2 for the subsample of the balanced panel of refugees who get asylum after the first municipalities’ quotas for that calendar year have been filled, making refugees’ own preferences for settlement less likely to influence municipality assignment.

25

Since the dependent variable is a dummy variable, in fact we estimate linear probability models. Our model specification uses dummy variables as control variables in order to satisfy the requirements of a saturated model. The saturated model with a discrete outcome will identify identical coefficient estimates and standard errors to a logit/probit model (Angrist 2001). 26 It is not possible to estimate a panel data model with an individual fixed effect, because for any given refugee, there is only one year of arrival and hence no time-variation in the model. Estimation of a panel data model with an individual random effect would rely on the strong assumption of no correlation between the individual random effect (capturing e.g. innate ability) and the independent variables (e.g. level of education). 27 Åslund and Rooth (2007) also provide instrumental variables (IV) estimates of the effect of current local labour market conditions on individual employment outcomes of refugees subject to the Whole of Sweden Strategy. In particular, they instrument the unemployment rate in the current municipality of residence by the unemployment rate in the municipality of assignment in the year of asylum. Consistency of the IV-estimate requires an additional identifying assumption relative to the reduced form (our baseline) model: The initial labour market conditions only affect the individual’s current labour market outcomes through its impact on the current labour market condition (the exclusion restriction). Scarring effects would render such an instrument invalid. The advantage of the reduced model is that it gives consistent estimates even if the exclusion restriction required for consistency of the IV-model is not satisfied.

11

IV. DATA The following section provides a description of the raw data and our sample selection criteria. Furthermore, we conduct an initial investigation of the dataset, focusing on the geographical dispersion of refugees across municipalities and investigation of whether (subsamples of) refugees subject to the dispersal policy were initially randomly distributed across municipalities, conditional on demographic characteristics. Finally, we describe the labour market attachment of refugees by years since asylum and present summary statistics of our samples.

IV.A. Data sources and sample selection The empirical analysis presented in this study is based on longitudinal administrative register data from Statistics Denmark for the years between 1999 and 2015. DIS has detailed information on granted residence permits from 1997 onwards, allowing us to perfectly identify refugees for the period of interest. Using a unique person identifier, it is possible to link the data from DIS with the Danish population register that contains demographic characteristics, such as gender, age or residence and other records maintained by Statistics Demark, like labour market status. The data from DIS does not contain exact information on the initial municipality of placement, instead, this is retrieved from the population registers. Particularly, it is possible to trace people’s municipality of residence, determined at the end of each year. We treat the first municipality registered as the municipality of assignment, if the refugee is recorded in the registries the year in which asylum was granted, or in the subsequent year.28 The administrative-territorial structure of Denmark has undergone a major structural reform from 2003 and onwards that culminated in the local and regional government reform of 2007. The reforms reduced the number of municipalities from 275 to 98 (LGDK, 2009). Throughout our analyses, we use post-reform municipalities. However, to do so we need to solve the data break resulting from the reform and assign a post-reform municipality identifier to pre-reform municipalities. This data break affects the municipality level variables: the municipality quotas, local labour market conditions and the municipality of assignment of the refugees. All pre-reform municipalities, except for 13, translate one-to-one to the new municipalities. For the municipalities that are split into 2 (in one case 3) municipalities, we assign the full pre-reform population to the post-reform municipality to which the majority of the pre-reform population belong post-reform.29 To increase efficiency of our estimates we wish to control for the level of education attained before immigration in our regression analyses. The reason is that the educational level of refugees before immigration can influence their labour market integration (for previous evidence see e.g. Damm 2009a). Information regarding educational attainment from abroad is generally obtained through surveys conducted by Statistics Denmark. In case of nonresponse, Statistics Denmark imputes the value, but in order to avoid endogeneity issues, we have excluded this information and consider the educational level unknown in those cases. Because educational attainment in the host country may be endogenously determined, we do not control for such education in our regressions.

28

A robustness check for 2008-2010 shows that this is a sensible assumption. Indeed, only 1.7-3.9% of assigned refugees moved from the initial municipality within the first 12 months of residence; the 1.7% refers to the relocation rate after one quarter, while the 3.9% refers to the relocation rate after three quarters (authors’ own construction using quarterly population registers available for 2008-2010). 29 We expect this inaccuracy to have little influence on the later investigation as only 2% of the national population lives in these thirteen municipalities and only 3% of the individuals granted asylum are allocated to those municipalities.

12

We extract four samples of refugees from the linked permit of residence and population registers. Each sample is in turn described below. The gross sample of refugees, or simply ‘the gross sample’ comprises the newly arrived adult refugees in the period 1999 to 2015. We focus on refugees arriving as adults in the period 1999 to 2010.30 Individuals who are not recorded in the population registry within a year after they obtain a residence permit are excluded from the gross sample. The gross sample has 12,692 individuals. From the gross sample we extract a balanced panel of household heads also referred to as ‘the balanced panel’. It consists of individuals who are observed for at least four years since assignment, who were in their working ages (aged 18 to 59) at the time of asylum and who are household heads. The balanced panel includes 8,479 individuals.31 The choice of four years stems from our genuine interest to analyse the effects of the initial labour market conditions after the three-year introduction program, without compromising the sample size.32 The age-restriction is imposed because our aim is to analyse the integration of refugees into the labour market. We limit the balanced panel to household heads (i.e. first-arrived adult) as they represent the main subjects of the dispersal policy, whose settlement other family members’ settlement depends on. If a married couple of refugees is granted asylum on the same date, we consider the husband to be the household head.33 In principle, the assignment to a municipality of the first family member will determine the assignment of close family members to the same municipality, since as described earlier, DIS does not split spouses and children even if municipality quotas are filled. However, later arrived family members may arrive after the household head has moved away from the municipality of assignment. Inclusion of such family members into our estimation sample would bias our results. If household heads were initially completely randomly distributed across the country, we would expect personal characteristics and characteristics of the municipality of assignment to be uncorrelated. If the individual educational attainment is correlated with municipality of assignment characteristics, it raises our concern that more able individuals have realized more favourable settlement conditions. Therefore, for each municipality of assignment characteristic that we observe, we investigate whether individual educational attainment is correlated with the municipality of assignment characteristics, controlling for individual demographic characteristics observed by DIS: age, indicators for being male, marital status, having children in different age groups, country of origin, year and month of asylum. We observe the following assigned municipality characteristics: the unemployment rate, unemployment rate among non-Western immigrants, employment rate, employment rate among non-Western immigrants, employment growth, population share, non-Western immigrants share, co-national share, three different measures of commuting distance to centre of local labour market, and the annual influx of assigned refugees per 1,000 inhabitants. We report the coefficient estimates of the level of education (10-12 years, more than 12 years, using below 10 years as the reference category) in Table A1 in the Appendix. The results for the balanced sample are shown in Panel A. We find that individuals with high education were significantly more likely to be assigned to a municipality with a higher local 30

We restrict the sample to refugees arriving before 2011 for two reasons. First, because we wish to extract a panel of refugees whom we can follow in the administrative registers for at least four years since asylum; given that 2015 is our last year of observation, this limits our sample to refugees arriving before 2012. Second, the number of refugees arriving in 2011 is unexpectedly low. As a consequence, the number of refugees in the 2011 cohort who arrived after the first 10 municipalities had their annual refugee quota filled was close to zero. 31 For a detailed description of the sample reduction after each sample selection criteria see Table B1 in the online Appendix. 32 Leaving the panel is caused by out-migration, but only 1% of the gross sample out-migrate within the four-year observation period. 33 The reasons for choosing the male partner are twofold. First, our data shows that female partners are unlikely to immigrate before the husband. Second, as the data description will show, in the context of non-Western immigrants, men have a higher labour market attachment than women.

13

employment rate. In Panels B-G, we report the coefficient estimates of the level of education in each regression for various subsamples of the balanced panel excluding household heads who receive asylum before respectively the first 4, 8, 9, 10, 11 and 14 municipalities have filled their annual refugee quota.34 For the subsamples drawn by using a lower cut-off than 10 full municipalities, we find that individuals with high education were significantly more likely to be assigned to a municipality with a higher local employment rate (significant at the 5 or 10% significance level). By contrast, for the subsample of household heads who get asylum after the first 10 municipalities within a given year have their quotas filled (Panel E), we find that educational groups have been randomly distributed across municipalities, conditional on personal demographic characteristics observed by DIS. The preferred subsample for estimation of the effects of initial labour market conditions would be the largest subsample that provide quasirandom allocation, i.e. passes the balancing test. Panels B-G show that the preferred subsample for estimation is therefore the subsample of the balanced panel of household heads who got asylum after the first 10 municipalities within a given year have their quotas filled. This selection criterion reduces our subsample to 4,282 household heads. Hereafter, we refer to this subsample as the subsample of the balanced panel of household heads or simply ‘the subsample’. Note that this selection criterion restricts our sample not only to individuals subject to the Danish Dispersal Policy, but also to those arriving after the first municipality quotas have been filled in order for refugees to be even less likely to influence municipality assignment. We report the full set of estimates from our balancing test for ‘the subsample’ in Table 1. Using a 5-percent significance level one or two demographic characteristics of the household head are correlated with a given municipality characteristic.35 Given that we are testing 6 individual characteristics against 12 municipality characteristics it is not surprising that we find some correlations of which the gender-variation is the most common. We will use all individual characteristics as control variables in the analysis.36 [Table 1. Balancing tests. Include around here] Had we restricted the sample to those refugees arriving in the very last month of the calendar year, it would yield a very small sample and imprecise estimates and, in years in which municipal quotas are never filled, refugee settlement may still be endogenous. Finally, we construct a dataset for analysis of the effects of local labour market conditions on the employment probability. The dataset consists of the subsample augmented with observations for spouses who also get asylum on the same date as the household head (extracted from the gross sample); inclusion of such spouses augments the subsample with observations for 814 individuals. Inclusion of such spouses into our estimation sample increases efficiency of the estimations and increases the external validity of our results by inclusion of more married female refugees into the sample. Henceforth, we refer to this fourth sample as the subsample of household heads and jointly 34

We ran the balancing tests for all subsamples of the balanced sample including the individuals that are granted refugee status once j municipalities had been filled, where j=1,2…20. 35 Male household heads are less likely to be assigned to a municipality with a high unemployment rate and more likely to be assigned to a municipality with a high employment rate and longer distance to centre of local labour market, while there are no gender differences related to other municipality characteristics. Older household heads are more likely to be assigned to locations with a relatively high employment rate, longer distance to centre of local labour market and a higher annual influx of assigned refugees, while age of the household head is not correlated with other municipal characteristics. Married household heads are slightly less likely to be assigned to a location with a relatively high employment growth, while having children aged 3-17 is only slightly negatively correlated with the co-national share, and slightly positively correlated with the annual influx of assigned refugees. 36 According to Pei, Pischke and Schwandt (2017) a generally more powerful way of testing the relationship is to use the proxy for the candidate confounder (in our case educational level at the time of asylum) on the left-hand side of the regression instead of the right-hand side. Therefore, we have conducted this balancing test as well and shown it in Table A2 in the Appendix. This test confirms that there is no correlation between individuals’ educational attainment (as measured by a dummy for having at least 10 years of education) and any of the 12 municipality characteristics.

14

arrived couples or simply ‘the estimation sample’; it has observations for 4,282 household heads and 814 spouses (arrived on the same date as the household head), summing to 5,096 individuals. For all samples, we merge the information on the first residence of all individuals with municipality time-series data constructed by the authors. Using primarily administrative registers from Statistics Denmark, we calculate the population share and non-Western immigrant share, co-national share and labour market characteristics for each municipality in the observation period. Another way of testing whether individuals in the estimation sample have been randomly assigned across locations, conditional on the few demographic characteristics known by DIS, is to test whether better educated individuals were more likely to realise their preferred location choice initially. To test this, we regress an indicator for having moved out of the assigned municipality within four years after assignment on individual educational attainment, controlling for demographic characteristics observed by DIS and demographic and labour market characteristics of the assigned municipality. The relocation rate is 17%. We report the coefficient estimates from estimation of the model as a linear probability model in Table 2.37 Irrespective of specification, we find that the relocation rate is not affected by the level of education of the individual. This result provides further empirical support to our identifying assumption that individuals in our estimation sample were initially randomly assigned to locations, conditional on a few demographic characteristics. [Include Table 2 around here]

IV.B. Summary statistics Tables B2-B3 in the online Appendix report summary statistics for refugees across our four samples: 1) the gross sample, 2) the balanced panel, 3) the subsample and 4) the estimation sample. Generally, refugees are often men travelling alone, while family reunified (arriving later) are more often women and children. Besides, in cases in which family members arrive at the same date, we consider the man as household head and thereby include him in the balanced panel and subsample. For these two reasons, men constitute 82% of the individuals in the balanced sample and the subsample against 66% in the gross sample and 69% in the estimation sample.38 Due to exclusion of refugees above age 60 from the estimation sample, the individuals in the estimation sample are slightly younger and marginally more likely to be married and have children than individuals in the gross sample. Information about immigrants’ education from abroad was gathered by Statistics Denmark less systematically between 2006 and 2016. As a result, the share of the samples with unknown education varies by year of arrival but remains stable across samples.

37

Coefficient estimates of the municipality of assignment characteristics included as additional explanatory variables show that refugees’ probability to relocate decreases with the local share of non-Western immigrants and local employment rate. Results using the local unemployment rate as the main local labour market characteristic instead of the local employment rate (Table A3 in the Appendix) confirm that poor local labour market conditions are a significant push factor. Our results are consistent with previous quasi-experimental evidence on push and pull factors in immigrant settlement (Åslund 2005; Damm 2009b; Damm 2014). 38 A comparison of the individual characteristics in the gross and estimation samples by gender shows that the exclusion of later arrived spouses from the estimation sample as expected makes a larger difference for women than men.

15

As the estimation sample consists of refugees arrived after the first municipality quotas have been filled and each year starts with new quotas in January, we find that none of the refugees in the estimation sample arrive in January and only few arrive in February, while this is fairly common in the gross sample. As expected, we also find a lower representation of individuals in the estimation sample arriving in years like 2002 and 2010, when the actual number of newly arrived refugees was lower than expected by the authorities the year before. On the contrary, a higher share of the sample is represented in years like 1999 and 2001, when the authorities underestimated the number of asylum applicants. This can also explain why nationalities that arrive in the early years, like Afghans and Iraqis, have a slightly higher representation in the estimation sample. All in all, the individual characteristics of refugees in the gross sample and the estimation sample are rather similar and the differences found are closely related to the selection criterion for inclusion. While refugees in sample 1 are initially distributed across 97 out the 98 municipalities, refugees in sample 2 are initially distributed across 95 municipalities. Refugees in sample 3 and 4 are assigned to 94 different municipalities which are on average slightly smaller, with a smaller concentration of non-Western immigrants and co-nationals, higher unemployment among non-Western immigrants and lower employment rates among non-Western immigrants.39 In order to test whether refugees who obtain asylum early in the calendar year were more likely to have their location preferences met, in Table 3 we compare municipality characteristics for those refugees who got asylum before the first 10 municipalities had filled their annual quota (i.e. the individuals in the balanced panel who are excluded from the subsample) (column 2) with those who got asylum after (the subsample) (column 3). In column 4, we report ttests for differences in mean values of municipality characteristics for the two groups. Our results indicate a tendency for refugees who got asylum early in the calendar year to be assigned to larger municipalities and to municipalities with larger shares of non-Western immigrants and co-nationals and more favourable local labour markets for nonWestern immigrants,40 mirroring the settlement pattern of refugees whose settlement is no longer restricted by a Spatial Dispersal Policy (Åslund 2005; Damm 2009b). The descriptive results in Table 3 thereby confirm our strategy of excluding the first arrived refugees from our estimation sample. [Table 3. Insert around here] Readers are referred to Table B5 in the online Appendix for summary statistics for local municipality characteristics for the period 1999 to 2014 and correlations between municipality characteristics. The table includes 1,568 observations corresponding to yearly information for each of the 98 municipalities over a period of 16 years. As expected, the highest correlation (0.74) is found between the employment and the unemployment rate. Table B6 in the online Appendix provides detailed information about the definition of each variable and the data source used to construct each variable.

Finally, we investigate to which extent local labour market conditions are persistent over time. We report the results in Table A4 in the Appendix for the local employment rate (Panel A) and the local unemployment 39

Table B4 in the online Appendix shows that the employment rate of refugees increases by years since asylum for both genders, but it differs greatly between men and women. For men in the gross sample it increases from 32% in year 2 to 44% in year 4 since asylum. For women it increases from 10% in year 2 to 22% in year 4 since asylum. 40 As for labour market characteristics, the test results show that refugee household heads who got asylum relatively early in the calendar year are significantly more likely to be assigned to municipalities characterized by lower distance to the centre of the local labour market and a somewhat higher employment rate for non-Western immigrants, lower unemployment rate for nonWestern immigrants, but also a lower employment growth and slightly lower general employment rate.

16

rate (Panel B). Over our observation period the correlation between the employment rate in year t and t+s, for s=2-4, is between 0.85-0.91; the correlation between the unemployment rate in year t and t+s, for s=2-4, is only between 0.28-0.55. Our results lend further support to using the local employment rate as our preferred measure of local labour demand.

V. RESULTS In this section, we present our baseline results regarding the effects of local labour demand on immigrants’ employment status in November for each year 2-4 years since asylum. These estimations are based on the balanced panel and the estimation sample described earlier. A strength of our research design is that we conduct the analyses for refugees who were subject to spatial dispersion upon asylum and estimate the effects of initial local labour demand in the municipality of assignment. To evaluate the direction of the bias due to location sorting, we also report the estimates of the current local labour demand 2-4 years since asylum. Finally, we report the results from a range of robustness checks such as estimated effects of alternative measures of local labour market demand, including unemployment-based measures, as well as results from heterogeneity analyses.

V.A. Baseline results We present our baseline results in Table 4. The first two panels report the statistical association between the individual’s current employment status (in Nov.) and the employment rate in the current municipality of residence, controlling for initial personal attributes and other characteristics of the current municipality of residence (Eq. 2), using the balanced panel (Panel A) and the estimation sample (Panel B). Recall that the balanced panel may include refugees who had their location wish fulfilled upon asylum. Therefore, in Panel B we present results from estimation of Eq. 2 using the estimation sample, that consists of household heads and jointly arrived couples who arrive after the first 10 municipality quotas are filled. In Panel C we use the estimation sample again, this time reporting the results from our preferred model (Eq. 1). We interpret the coefficient estimates in Panel C as the estimated effects of the initial local employment rate. The estimated models include the following control variables: age, indicators for gender, marital status, children aged 0-2, children aged 3-17, educational attainment, country of origin, year of asylum, month of asylum, years since asylum as well as the municipality’s share of the total population in Denmark and the share of non-Western immigrants in the municipality. [Table 4. Baseline results. Include around here.] Focusing on the estimated coefficient of the current local employment rate in Panel A, we find that the estimated effects are all highly significant and (as expected) positive. Our estimates suggest that a one percentage point increase in the current local employment rate is associated with a 0.7-0.8 percentage point increase in refugee employment, 24 years after asylum. However, as the sample used in Panel A includes refugees arriving before the first ten municipality quotas are filled, we cannot rule out that (the most employable among) the first arrived refugees within a calendar year have self-selected into municipalities with a more favourable local labour market. Consistent with this hypothesis, the estimated effects are somewhat lower, 0.6-0.7, in Panel B which uses the subsample arrived after filling of the first 10 municipalities in the year. However, since we are measuring the effect of the current local employment rate in the current municipality of residence in Panel B, the estimates may still be biased to the extent that initially assigned individuals in the subsample have subsequently self-selected into locations. In fact, 17% of refugees in the estimation sample have relocated away from the municipality of assignment within four years after 17

asylum, and their decision to move is significantly affected by initial local labour market conditions (Table 2). Bias due to subsequent location sorting is ruled out in Panel C where we estimate the (ITT) effects of the initial local employment rate on the refugees’ employment. The high time persistency of the local employment rate (Table A4 in the Appendix) rules out bias due to omission of the current employment rate in the assigned municipality. As expected, the estimated effects decrease to 0.5-0.6, but are still highly significant.41 The lower magnitude of the estimates in Panel C compared to A and B provides quasi-experimental evidence of positive self-selection of refugees into areas with more favourable labour market conditions. Failure to account for such location sorting leads to an upward bias of the estimates. Hence, we conclude that being allocated to a municipality with a 1 percentage point higher employment rate increases the employment probability of refugees by 0.5-0.6 percentage points, 2-4 years after asylum. As the employment level for newly arrived refugees is low, this corresponds to an increase in refugees’ employment rate after 2, 3 and 4 years since asylum of 2.3%, 2.0% and 1.5%.42 43

V.B. Robustness and heterogeneity analyses In the following we conduct series of robustness checks and heterogeneity analyses. Our first robustness check investigates how sensitive the estimated intent-to-treat estimates of the initial local employment rate in Table 4, Panel C, are with respect to the choice of cut-off with respect to the number of municipalities that had filled their annual refugee quota. We do so by estimating Eq. 1 for various cut-offs for exclusion of individuals in the balanced panel from the estimation sample, from no exclusion of individuals (cut-off of 0 municipalities with full annual refugee quota) to exclusion of individuals who have arrived before the first 14 municipalities have filled their annual refugee quota. We report the estimates of Eq. 1 in Table A6 in the Appendix for 0 full municipalities (Panel A), first 4 municipalities full (Panel B), first 8 municipalities full (Panel C), first 9 municipalities full (Panel D), first 10 municipalities full (Panel E), first 11 municipalities full (Panel G), first 14 municipalities full (Panel H). For comparison, our baseline estimates in Table 4, Panel C, are repeated in Table A6, Panel F. Our point estimates are not sensitive to whether we use a cut-off of 4, 8, 9, 10 or 11 municipalities with full annual refugee quota, but they are lower than the estimates for the balanced panel (Panel A). Using a cut-off of 14 municipalities with full annual quota slightly decreases the point estimates, while reducing the precision due to a 50% drop in the number of observations; however, the point estimates are not significantly different from the baseline estimates. We conclude that our

41

Given the combination of the low rate of geographical mobility after municipal assignment and the high time persistency of the local employment rate, we cannot distinguish empirically between two alternative interpretations of our results: (ITT) effects of the current local employment rate versus scarring effects. As for geographical lock-in effects, they may be substantial. Even though the local employment rate is a statistically significant push factor (Table 2), a standard deviation decrease in the local employment rate (4.61 percentage points) only increases the probability of having moved out of the assigned municipality within four years since assignment by 2.6% (calculated as 4.61 times the point estimate of -0.0061*100%). 42 Since the average local employment rate is high, a 1 percentage point change in local labour employment is close to a 1 percent change. Hence, the elasticity is close to the semi-elasticity: 1.8%, 1.5% and 1.1%, respectively, 2, 3 and 4 years since asylum. 43 In Table A5 in the Appendix we report the estimated effects of all three characteristics of the assigned municipality at arrival in our regressions (employment rate, population share, non-Western immigrant share). The reason for also reporting estimated effects of the population share and non-Western immigrant share is that it allows us to compare our results with Damm and Rosholm (2010). They find a negative effect of the population size and immigrant share on the hazard rate into first job of refugees subject to the Danish Spatial Dispersal Policy in place from 1986-1998. By contrast, our findings for refugees subject to the Danish Spatial Dispersal Policy 1999-2010 show that, ceteris paribus, employment of refugees was unaffected by whether they were assigned to a large city and to a location with a high share of non-Western immigrants in the population. Possible explanations include differences in model specifications (mixed proportional hazard model versus linear regressions using cross sectional data different functional form of the explanatory variables), sample period (1986-1998 versus 1999-2010) and sample differences (e.g. municipalities with zero annual refugee quota excluded from our sample).

18

baseline results are robust to a wide range of alternative cut-offs for inclusion in the estimation sample. For the remainder of the analysis we use the ‘estimation sample’ defined in Section IV. In our second robustness check, we estimate the effects of four alternative measures of local labour demand using our preferred specification (Eq. 1). The five different measures are: the unemployment rate (Panel B), the employment rate among non-Western immigrants (Panel C), the unemployment rate among non-Western immigrants (Panel D) and finally employment growth (Panel E). For comparison, in Table 5, Panel A, we repeat our baseline results using our preferred measure of the local labour demand, the employment rate, from Table 4, Panel C. The results show that, as expected, the coefficient estimates of the initial employment rate of non-Western immigrants are positive, and significantly so in year 2 and 3 since asylum. The coefficients are lower in magnitude than our baseline estimates. The coefficients of the initial unemployment rates are as expected negative and significant in all years, though only at the 10-percent level in year 2 since asylum. Measuring local labour demand by the unemployment rate of non-Western immigrants instead of the general unemployment rate results in smaller magnitudes of the estimates, but significant and negative effects as expected. The last measure of local labour demand is employment growth. Unlike the other labour demand measures the initial employment growth has an insignificant effect on the individual employment status of refugees in year 2-4 since asylum, which suggests that local employment growth is not of significance for refugees’ employment. The three other measures are in line with our baseline results, but in some cases estimated with lower precision. [Table 5. Include around here.] Next, we check whether our estimated effects of the initial local labour demand 2-4 years since asylum (in Table 5, Panels A and C) are robust to inclusion of other local characteristics, like job search networks and commuting costs. In other words, we test whether our baseline models adequately control for correlated effects. We include in turn the following additional controls: Commuting time to the centre of the commuting area (using public transportation and car), commuting distance, share of co-nationals and commuting area fixed effects. We use the definition of commuting areas by Statistics Denmark (2016). According to Statistics Denmark, there were 29 commuting areas in Denmark in 2014. Results from estimation of the model in Eq. (1) with these additional controls are reported in Table 6, in Panel A for the initial local employment rate and in Panel B for the initial local unemployment rate. The estimates of both measures are very robust to the inclusion of share of co-nationals, commuting time and distance. None of the estimates of additional observed municipality characteristics are significant at a conventional 5 percent significance level. Our results are slightly more mixed, when introducing commuting area fixed effects and thereby controlling for all variation between the 29 commuting areas in Denmark. The estimated effect of the local unemployment rate is almost halved in magnitude (-0.7 pp.) and insignificant, while the estimated effect of the local employment rate is less affected (0.1 pp.), but the estimate is now insignificant due to larger standard errors, which is not very surprising given the specification.44 45

44

Åslund and Roth (2007) also mention that they have experimented with a specification including both regional and cohort/time fixed effects. They conclude that the differences in within-region variation in unemployment over time appear to be too small to identify the effects of initial conditions. 45 We have also investigated the idea of estimating the effects of the employment rate of co-nationals who are living in the municipality of assignment at the time at which individual i is assigned to the municipality using only within-municipality variation, that is, including municipality of assignment fixed effects into Eq. 1. Unfortunately, there is insufficient time-variation in the employment rate of co-nationals at the municipal level for that specification. As shown in Damm (2014) estimation of the effects of the local employment rate of co-nationals exploiting only within-municipality variation requires neighbourhood data that currently only exist for Denmark for the period 1985-2004 (see Damm and Schultz-Nielsen 2008). Moreover, such neighbourhood effects analysis is beyond the scope of our study.

19

[Table 6. Include around here.] One could also argue that the local labour market conditions for low-skilled would be a relevant measure since many non-Western immigrants work in such positions. Therefore, we conduct a robustness check in which we measure local labour demand for low-skilled by the local employment rates and local unemployment rates among low-skilled individuals in Denmark. The estimated effects of initial local labour market conditions for low-skilled are reported in Table 7. The results show that when using employment rates for low-skilled as local labour market measure we find highly significant estimates that are similar to the baseline estimates in Table 4, Panel C. Turning to the estimated employment effect of the initial local unemployment rate for low-skilled, the estimates (-0.4 to -1.0 pp.) are highly significant, but somewhat lower in magnitude than in Table 5, Panel B. These results support our main findings. [Table 7. Include table around here.] Finally, we have conducted three heterogeneity analyses, the first is related to gender, the second to skill level and the third to period of arrival. [Table 8. Include table around here.] Table 8, Panels A and B, reports the results from Eq. 1, extended with an interaction term between the measure of labour market condition and a dummy for female, hereby allowing the estimated effect of initial local labour market conditions to differ by gender; as measure of local labour conditions we use the local employment rate (Panel A) and the local unemployment rate (Panel B). The main effect is significant in both models and the coefficient estimate is very similar to the baseline estimates in Table 5, Panels A and C, whereas the estimated effect of the interaction term is insignificant in both models. This means that the estimated effect of the local employment rate (local unemployment rate) does not vary by gender. The estimated effect of a one percentage point change in the local employment rate (local unemployment rate) is 0.6 percentage points (-1.9 percentage points) on the employment probability of both male and female refugees 4 years after asylum. However, in percentages this change is substantially larger for female than male refugees. Hence, the elasticity of the local employment rate (local unemployment rate) is 2.4 (-0.35) for females against 1.1 (-0.18) for males and thereby much larger in magnitude for females. The reason is that the employment rate of female refugees 4 years after asylum is only 21% compared to 44% for male refugees. In Table 8 we also report the estimates of Eq. 1, extending the model with an interaction term between our measure of the local labour market conditions and a dummy for having more than 12 years of education (high-skilled). As measure of local labour market conditions, we use the local employment rate (Panel C) and the local unemployment rate (Panel D). The main effects remain similar to the baseline estimates in Table 4, Panel C, and Table 5, Panel D, while the interaction terms are insignificant. Note, however, that four years after arrival the interaction term for highskilled individuals has the opposite sign, but similar magnitude as the main effect. We interpret it as an imprecisely estimated zero effect of local labour market conditions at arrival four years after arrival for high-skilled immigrants. By contrast, employment of low-skilled immigrants remains sensitive to local labour market conditions at arrival up to four years since arrival. Finally, in Table A7 in the Appendix we report separate estimates of the initial local employment rate on individual employment chances 2-4 years since asylum (Eq. 1) after splitting the estimation sample into two: individuals arriving 1999-2002, and individuals arriving after 2002. Our heterogeneity checks show that our results are driven by the first four cohorts of refugees in our sample (1999-2002), which constitute 70% of our sample. We can rule out that it is due to cohort differences in the relocation rate from the municipality of assignment. Instead, we speculate that it is due to 20

the lower time persistency of the local employment rate in the second part of the period as shown in Table A4 in the Appendix (Panels C and D).

VI. DISCUSSION AND CONCLUSIONS Our results show substantial gender differences in the speed of labour market integration of refugees who were subject to the Danish Spatial Dispersal Policy in place from 1999 until 2016. While the employment rate of male refugees increased during the first four years after asylum to around 44%, the employment rate of female refugees increased more slowly to reach only 22% four years after asylum. Most importantly, our study provides quasi-experimental evidence that refugees who were subject to the Danish Spatial Dispersal Policy in place from 1999 until 2016, ceteris paribus, had a higher employment probability if they were assigned to a municipality with favourable local labour market conditions, as measured by the local employment rate or the unemployment rate. The effect of the local labour demand is economically significant. Initial residence in a municipality with a one percentage point higher employment rate increases the employment rate of refugees by 0.50.6 percentage points (corresponding to elasticities of 1.1-1.8) two to four years after asylum in Denmark. A one percentage point higher local employment rate increases the employment probability of male and female refugees by the same amount in percentage points, but since the employment rate among female refugees is substantially lower than that of male refugees, the estimated effect of a one percentage point increase in the local employment rate is substantially larger for female refugees in percentage terms. If we instead measure the local labour demand by the local unemployment rate, we estimate that a one percentage point increase in the local unemployment rate in the municipality of assignment, decreases the individual employment probability of refugees by 0.9-1.7 percentage points (elasticities -0.16-(-0.19)) two to four years after asylum in Denmark. Our study provides quasi-experimental evidence that more able refugees self-select into locations with favourable employment conditions. Failure to account for location sorting of immigrants leads to overestimation of the sensitivity of immigrant employment to local employment conditions. This result is in accordance with Åslund and Rooth (2007). In other words, their results generalise to another Scandinavian country. Our study further shows how researchers can make better use of micro data for refugees subject to a Spatial Dispersal Policy due to municipal quotas, including the ‘Whole of Sweden’ Strategy used in e.g. Edin et al. (2003) and Åslund and Rooth (2007) and the Danish Spatial Dispersal Policy in place since 1999. In future research, instead of using the full population of refugees subject to the Spatial Dispersal Policy, consistency of estimated effects of location characteristics can be increased by extracting the subsample of refugees who got asylum after a substantial share of municipalities had reached their annual full capacity. Our findings of significant effects of the local labour market conditions support the reform of the Danish Spatial Dispersal Policy in July 2016. Since the reform, refugee settlement in a municipality with good employment prospects should be given important consideration.46 More generally, our results provide quasi-experimental evidence that immigrant employment is sensitive to labour market conditions in the initial location and highlight the importance of carefully designing refugee allocation policies.

46

Bekendtgørelse om boligplacering af flygtninge, BEK number 243, dated the 21st of March 2018, (or ”Boligbekendtgørelsen” for short), Section 12.

21

Appendix A. How are municipality quotas calculated? The annual quota is based on a calculation where each municipality’s share (sharemt) is equal to the municipalities share of 47 the total population times two minus the share of non-EU citizens living in the municipality:

 pop sharemt  2*  n mt   popmt  m1

  im    n mt    immt   m1

 ,  

where popmt is population in municipality m in year t, immt is non-EU citizens in municipality m in year t, n is the number of municipalities in Denmark. A preliminary municipality quota (quota_pmt) is calculated as the municipality share of the total sum on country level of refugees expecting to arrive (landstal) in year t and family reunified non-EU citizens (fam) arriving last year (t-1). From this quota is withdrawn the number of family reunified non-EU citizens arrived in the municipality last year (fammt-1).

quota _ pmt  sharemt (landstalt  famt 1 )  fammt 1 If the preliminary quota is negative it is set to zero.

quota _ pmt  0 if quota _ pmt  0 The proportion of expected refugees on country level not included in the overall sum of positive municipality quotas are distributed between municipalities in accordance with the preliminary quotas. The final municipality quotas (quotamt) are found as:

  landstalt  quotamt  quota _ pmt *  n   quota _ pmt   m1  A comparison of the actual municipality quotas and calculated quotas based on the above-mentioned formula for the last sample year (2010) is shown in Figure A1. The figure shows a close relationship between the actual and calculated quotas in the 98 municipalities. The correlation between the two quota-measures is 0.947. Please note that the remaining difference may be due to measurement error in the calculation since the calculated quotas rest on population data by January 1st 2010, while the actual quotas (announced October 2009) rest on earlier information. Source: “Order # 800 of 24/08/2000”, “Order # 1169 of 17/12/2002”, “Order # 914 of 12/11/2003” and “Order # 50 of 18/01/2008”.

References Algan, Y., C. Dustmann, A. Glitz and A. Manning (2010), “The Economic Situation of First and Second Generation Immigrants in France, Germany and the United Kingdom”, Economic Journal, 120(Feb.), F4-F30. 47

More precisely, ‘non-EU citizens’ here include all citizens from outside the EU/EEA and the Nordic countries.

22

Angrist, J. (2001), “Estimation of Limited Dependent Variable Models with Dummy Endogenous Regressors”, Journal of Business and Economic Statistics, 19(1), 2-28. Barth, E., B. Bratsberg and O. Raaum (2004), “Identifying Earnings Assimilation of Immigrants under Changing Macroeconomic Conditions”, Scandinavian Journal of Economics, 106(1), 1-22. Bauer, T.K., M. Lofstrom and K.F. Zimmermann (2000), “Immigration Policy, Assimilation of Immigrants and Natives' Sentiments towards Immigrants: Evidence from 12 OECD-Countries”, IZA Discussion Paper, 187. Beaman, L.A. (2012), “Social Networks and the Dynamics of Labour Market Outcomes: Evidence from Refugees Resettled in the U.S.”, Review of Economic Studies, 79, 128-161. Becker, S.O. and A. Ferrara (2019), “Consequences of forced migration: A survey of recent findings”, Labour Economics, 59, 1-16. Berman, E., J. Bound and S. Machin (1998), “Implications of Skill-Biased Technological Change: International Evidence”, Quarterly Journal of Economics, 113(4), 1245-1279. Bertrand, M. and S. Mullainathan (2004), “Are Emily and Greg More Employable than Lakisha and Jamal? A Field Experiment on Labour Market Discrimination”, American Economic Review, 94(4), 991-1013.

Borjas, G.J. (1985), “Assimilation, Changes in Cohort Quality and the earnings of immigrants”, Journal of Labor Economics, 3(4), 463–489. Borjas, G.J. (1995), “Assimilation, Changes in Cohort Quality revisited: what happened to earnings in the 1980s?” Journal of Labor Economics, 13(2), 201–245. Bratsberg, B., O. Raaum and K. Røed (2010), “When Minority Labor Migrants Meet the Welfare State”, Journal of Labor Economics, 22(3), 633-676. Bratsberg, B., O. Raaum and K. Røed (2014), “Immigrants, Labour Market Performance and Social Insurance”, Economic Journal 124, 644–683. Bratsberg, B., O. Raaum and K. Røed (2017), “Immigrant labor market integration across admission classes”, Nordic Economic Policy Review, 17-54. Card, D. and J. E. DiNardo (2002), “Skill-Biased Technological Change and Rising Wage Inequality: Some Problems and Puzzles”, Journal of Labor Economics, 20(4), 733-783. Carlsson, M. and D-O. Rooth, (2007), “Evidence of ethnic discrimination in the Swedish labor market using experimental data”, Labour Economics, 14(4), 716–729. Chiswick, B.R. (1978), “The Effect of Americanization on the Earnings of Foreign-born Men, Journal of Political Economy, 86(5), 897-921. Cortes, K.E. (2004), “Are Refugees Different from Economic Immigrants? Some Empirical Evidence on the Heterogeneity of Immigrant Groups in the United States”, Review of Economics and Statistics 86(2), 465–480. Damm, A.P. (2005), “The Danish Dispersal Policy on Refugee Immigrants 1986-1998: A Natural Experiment?”, Aarhus School of Business, Department of Economics WP 05-3.

23

Damm, A. P. and M. L. Schultz-Nielsen, (2008), “Danish Neighbourhoods: Construction and Relevance for Measurement of Residential Segregation”, Danish Journal of Economics (Nationaløkonomisk Tidsskrift), 146(3): 241262. Damm, A.P. (2009a), “Ethnic Enclaves and Immigrant Labor Market Outcomes: Quasi-Experimental Evidence”, Journal of Labor Economics, 27(2), 281-314. Damm, A. P., (2009b), “Determinants of Recent Immigrants’ Location Choices: Quasi-Experimental Evidence”, Journal of Population Economics, 22(1), 145-174. Damm, A.P. (2014), “Neighborhood Quality and Labor Market Outcomes: Evidence from Quasi-Random Neighborhood Assignment of Immigrants”, Journal of Urban Economics, 79, 139-166. Damm, A.P. and C. Dustmann, (2014), “Does Growing Up in a High Crime Neighborhood affect Youth Criminal Behavior?” American Economic Review, 104(6), 1806-1832. Damm, A.P. and M. Rosholm (2010), “Employment Effects of Spatial Dispersal of Refugees”, Review of Economics of the Household, 8(1), 105-146. Dustmann, C. (1993), “Earnings Adjustment of Temporary Migrants”, Journal of Population Economics, 6, 153-168. Dustmann, C., F. Fasani, T. Frattini, L. Minale and U. Schönberg, (2017) On the economics and politics of refugee migration. Economic Policy, 32, 497–550. Dustmann, C., Glitz, A., and T. Vogel. (2010), “Employment, Wages, and the Economic Cycle: Differences between Immigrants and Natives”, European Economic Review, 54(1), 1–17. Dustmann, C. and J-P Görlach. (2015), “Selective out-migration and the estimation of immigrants’ earnings profiles”, in B.R. Chiswick and P.W. Miller (eds), Handbook of the Economics of International Migration, 1A, 489–533, Elsevier.

Dustmann, C., K. Vasiljeva and A.P. Damm (2019), “Refugee Migration and Electoral Outcomes”, Review of Economic Studies, 86, 2035-2091. Edin, P-A, P. Fredriksson and O. Åslund (2003), “Ethnic Enclaves and the Economic Success of Immigrants - Evidence from a Natural Experiment”, Quarterly Journal of Economics, 118, 329-357. Edin, P-A, P. Fredriksson and O. Åslund (2004), “Settlement policies and the economic success of immigrants”, Journal of Population Economics, 17(1), 133-155. Esmail, A. and S. Everington (1997), “Asian Doctors Are Still Being Discriminated Against”, British Medical Journal, 314(May), 1619. Fasani, F., T. Frattini and L. Minale (2018), “(The Struggle for) Refugee Integration into the Labour Market: Evidence from Europe”, IZA Discussion Paper 11333. Godøy A., (2017), “Local labor markets and earnings of refugee immigrants”, Empirical Economics, 52(1), 31-58. Goos, M., A. Manning and A. Salomons (2014), “Explaining Job Polarization: Routine-Biased Technological Change and Offshoring”, American Economic Review, 104(8), 2509-26.

24

Grossman, G.M. and E. Rossi-Hansberg (2008), “Trading Tasks: A Simple Theory of Offshoring”, American Economic Review, 98(5), 1978-1997. Hoynes, H.W. (2000), “Local Labor Markets and Welfare Spells: Do Demand Conditions Matter?, Review of Economics and Statistics, vol. 82(3), 351-368. Husted, L., H.S. Nielsen, M. Rosholm and N. Smith (2001), “Employment and Wage Assimilation of Male FirstGeneration Immigrants in Denmark”, International Journal of Manpower, 22, 39–68. Hvidtfeldt, C. and M.L. Schultz-Nielsen (2018), “Refugees and Asylum seekers in Denmark 1992-2016”, The ROCKWOOL Foundation Research Unit, Study Paper 133. Hvidtfeldt, C., M.L. Schultz-Nielsen, E. Tekin and M. Fosgerau (2018), “An estimation of the effect of waiting time in the Danish asylum system on post-resettlement employment among refugees: Separating the pure delay effect from the effects of conditions under which refugees are waiting”. PLOS ONE 13(11), e0206737. https://doi.org./10.1371/ journal.pone.0206737. Katz, L.F. and K.M. Murphy (1992), “Changes in Relative Wages, 1963-1987: Supply and Demand factors”, Quarterly Journal of Economics, 107(1), 35-78. Kennan, J. (2013), “Open borders”, Review of Economic Dynamics, 16(2), L1–13. Lalonde, R.J. and R.H. Topel, (1992), “The Assimilation of Immigrants in the U.S. Labor Market,” In: Borjas, George J. and Richard B. Freeman (eds.) “Immigration and the Workforce: The Economic Consequences for the United States and Source Areas”, Chicago and London: University of Chicago Press, 67-92. LGDK (2009), “The Danish Local Government System”. Copenhagen: The Danish Local Government System (LGDK). Lubotsky, D. (2007), “Chutes or Ladders? A Longitudinal Analysis of Immigrant Earnings”, Journal of Political Economy, 115(5), 820-867. Moore, M.P. and P. Ranjan (2005), “Globalisation vs Skill-Biased Technological Change: Implications for Unemployment and Wage Inequality”, Economic Journal, 115(April), 391-422. Munshi, K. (2003), “Networks in the Modern Economy: Mexican Migrants in the U.S. Labor Market”, Quarterly Journal of Economics, 118(2), 549-599. Nielsen, C.P. and K.B. Jensen (2006), “The Danish Integration Act’s Significance for the Settlement Patterns of Refugees”, Copenhagen: Institute for Local Government Studies. Pei, Z., J-S. Pischke, and H. Schwandt (2017), “Poorly Measured Confounders Are More Useful On the Left than on the Right”, NBER WP 23232. Riach, P.A. and J. Rich (1991), “Testing for racial discrimination in the labour market”, Cambridge Journal of Economics, 15(3), 239-256. Ruiz, I. and C. Vargas-Silva (2018), “Differences in labour market outcomes between natives, refugees and other migrants in the UK”, Journal of Economic Geography, 18(4), 855-885.

25

Sarvimäki, M. (2011), “Assimilation to a Welfare State: Labor Market Performance and Use of Social Benefits by Immigrants to Finland”, Scandinavian Journal of Economics 113(3), 665–688. Sarvimäkki, M. (2017), “Labour market integration of refugees in Finland”, Nordic Economic Policy Review, 91-114. Schultz-Nielsen, M.L. (2016), ”Arbejdsmarkedstilknytningen for flygtninge og indvandrere”. The ROCKWOOL Foundation Research Unit. The University Press of Southern Denmark, Odense. Schultz-Nielsen, M.L. (2017), “Labour market integration of refugees in Denmark”, Nordic Economic Policy Review, 5590. Schultz-Nielsen, M.L. (2019), ”Immigrants’ assimilation depending on residence type”. The ROCKWOOL Foundation Research Unit, Study Paper 140. Smith, N., P.J. Pedersen, S. Pedersen, and M.L. Schultz-Nielsen, (2003), “Fra mangel på arbejde til mangel på arbejdskraft”. Copenhagen: Spektrum. Statistics Denmark (2016), ”Færre og større pendlingsområder”. DST Analyse 2016:23. Statistics Denmark (2018), “Indvandrere i Danmark 2018”. Copenhagen: Statistics Denmark. Storesletten, K. (2000), “Sustaining fiscal policy through immigration”, Journal of Political Economy 108, 300–23. Åslund, O. (2005), “Now and forever? Initial and subsequent location choices of immigrants”, Regional Science and Urban Economics, 35,141–165. Åslund, O., A. Forslund, and L. Lijleberg (2017), “Labour market entry of non-labour migrants – Swedish evidence”, Nordic Economic Policy Review, 115-158. Åslund, O., Hensvik, L. and Skans, O. N. (2014), “Seeking similarity: How immigrants and natives manage in the labor market”, Journal of Labor Economics, 32(3), 405-441. Åslund, O., and D-O. Rooth (2007), “Do Where and When Matter? Initial Labour Market Conditions and Immigrant Earnings”, Economic Journal, 117(518), 422-448.

26

Table 1. Balancing tests: Assigned municipality characteristics and individual characteristics of assignees (household heads). Dependent variable: Assigned municipality characteristics at arrival Unempl Unempl Emplo Emplo Emplo Popu Non- Comm Com Dist CoAnnu oyment oyment yment yment yment latio West uting mutin anc natio al rate rate of rate rate growt n ern distanc g e to nal influx nonof h share immi e to dista cent share of Wester nongrant center nce er assig n Weste s of local to of ned immigr rn share labour cente loca refug ants immig market r of l ees rants by local labo per public labou r 1,000 transp r mar inhab ortatio mark ket itants n et by car (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Explanatory variables: Individual characteristics at arrival Educational attainment (reference category: 0-9 years): 0.0021 0.015 0.01 0.040 0.59 0.005 0.020 10-12 years 3 0.130 0.130 0.114 2 49 0 0.750 0.593 4 50 7 (0.0644 (0.160 (0.343 (0.033 (0.04 (0.04 (0.79 (0.6 (0.00 (0.01 ) (0.325) ) ) 1) 02) 80) (1.125) 3) 69) 603) 77) More than 0.052 0.03 0.077 0.64 0.005 0.033 12 years -0.0429 -0.0249 0.231 0.129 1 42 0 -0.967 0.907 1 36 3 (0.0731 (0.186 (0.407 (0.037 (0.03 (0.04 (0.91 (0.7 (0.00 (0.02 ) (0.371) ) ) 4) 65) 90) (1.332) 7) 69) 603) 11) 0.160* 0.564* 0.570 0.684 0.024 0.07 0.007 1.718 1.64 0.003 0.004 Male ** * *** ** 0 04* 78 1.995* ** 3** 86 87 (0.0587 (0.155 (0.331 (0.033 (0.03 (0.04 (0.78 (0.6 (0.00 (0.01 ) (0.285) ) ) 1) 76) 71) (1.094) 5) 68) 676) 71) 0.0028 0.0061 0.012 0.023 0.001 0.00 0.003 0.0936 0.061 0.04 0.000 0.001 Age 1 3 9* 5 50 142 44 * 0* 73 128 96** (0.0 (0.0026 (0.0132 (0.006 (0.015 (0.001 (0.00 (0.00 (0.053 (0.03 310 (0.00 (0.00 0) ) 98) 2) 45) 19) 227) 0) 68) ) 0321) 0819) 0.067 0.02 0.079 0.84 0.010 0.018 Married 0.0402 0.0561 -0.179 -0.283 9** 79 3* -1.509 0.882 5 7* 6 (0.0563 (0.148 (0.314 (0.031 (0.03 (0.04 (0.74 (0.6 (0.00 (0.01 ) (0.272) ) ) 4) 89) 64) (1.090) 7) 30) 637) 64) Children 0.061 0.004 0.01 0.035 0.96 0.010 0.014 aged 0-2 -0.0364 0.273 0.156 5 70 6 7 -0.657 1.042 8 2 3 27

(0.0693 ) Children aged 3-17

Country of origin F.E. Year of arrival F.E. Month of arrival F.E. R2 Number of municipalitie s Number of individuals

0.0442 (0.0559 )

(0.276)

(0.178 ) 0.057 8 (0.142 )

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

0.286

0.122

0.351

0.712

0.136

0.075

Yes 0.06 7

Yes

0.343

Yes 0.05 5

0.430

0.374

(0.323) 0.0043 4

(0.360 ) 0.091 5 (0.304 )

(0.036 5) 0.001 44 (0.030 5)

(0.03 67) 0.05 79* (0.03 16)

(0.04 78) 0.075 7* (0.04 14)

(1.369)

1.318 (1.066)

(0.90 3)

(0.7 60)

0.629 (0.73 1)

0.067

0.63 2 (0.6 16)

(0.00 655) 0.015 7*** (0.00 571)

(0.01 89) 0.036 4** (0.01 66)

94 4,282

Source: Administrative register information from Statistics Denmark from 1999-2016. Note: ***: p<0.01, **: p<0.05, *: p<0.1. Robust standard errors in parentheses. The sample is the subsample of refugee household heads who got asylum during 1999-2010, who were observed in the first four years since arrival and who arrived after the first 10 municipalities had been filled in their calendar year of arrival. Reported coefficients are based on linear regressions of municipality characteristics in the year of arrival (asylum) on individual characteristics in the year of arrival. Additional control: Dummy for missing information on educational attainment at arrival.

28

Table 2. Effects of assigned municipality characteristics at arrival on the individual's probability of relocation Dependent variable: Moved out of the assigned municipality within four years since arrival (1) (2) (3) (4) (5) (6) Individual characteritistics at arrival Educational attainment (reference category: 0-9 years) 10-12 years 0.00197 0.00158 0.00125 0.00118 0.00127 0.00126 (0.0143)

(0.0142)

(0.0142)

(0.0142)

(0.0142)

0.0155

0.0157

0.0150

0.0152

0.0150

0.0150

(0.0172)

(0.0172)

(0.0173)

(0.0173)

(0.0174)

(0.0174)

0.0368***

0.0366***

0.0364***

0.0365***

0.0364***

0.0364***

(0.0101)

(0.0101)

(0.00999)

(0.00999)

(0.00999)

(0.00999)

Age

-0.00393***

-0.00392***

-0.00389***

-0.00391***

-0.00389***

-0.00389***

(0.000696)

(0.000695)

(0.000691)

(0.000695)

(0.000695)

(0.000696)

Married

-0.0399***

-0.0405***

-0.0408***

-0.0406***

-0.0408***

-0.0408***

(0.0151)

(0.0151)

(0.0151)

(0.0151)

(0.0151)

(0.0151)

Children aged 0-2

-0.0469***

-0.0460***

-0.0459***

-0.0460***

-0.0459***

-0.0459***

(0.0146)

(0.0146)

(0.0146)

(0.0147)

(0.0147)

(0.0147)

Children aged 3-17

-0.107***

-0.107***

-0.107***

-0.107***

-0.107***

-0.107***

(0.0138)

(0.0139)

(0.0138)

(0.0138)

(0.0137)

(0.0138)

-0.00610***

-0.00620***

-0.00396*

-0.00393*

-0.00393*

-0.00394*

(0.00177)

(0.00174)

(0.00227)

(0.00228)

(0.00229)

(0.00230)

More than 12 years Male

(0.0142)

Characteristics of assigned municipality at arrival

Employment rate at arrival (year t) Employment rate of non-Western immigrants at arrival (year t) Population share at arrival (year t) Non-Western immigrant share at arrival (year t) Co-national share at arrival (year t)

-0.00169

-0.00173

-0.00168

-0.00169

(0.00118)

(0.00118)

(0.00118)

(0.00118)

-0.00884

-0.00724

-0.00813

-0.00702

-0.00846

-0.00835

(0.01123) -0.04646*** (0.00855)

(0.01127) -0.04411*** (0.00862)

-0.01176 -0.04231*** (0.00878)

-0.01224 -0.04208*** (0.00874)

(0.01293) -0.04232*** (0.00878)

(0.01287) -0.04231*** (0.00878)

-0.0620

-0.0688

-0.0689

-0.0689

-0.0689

(0.0475)

(0.0471)

(0.0470) 0.000121 (0.000274)

(0.0468)

(0.0467)

Commuting time using public transportation Commuting time by car

-4.85e-05 (0.000450)

Commuting distance

-4.00e-05 (0.000545)

R2 Number of observations

0.118

0.118

0.119

0.119

0.119

0.119

5,096

Source: Administrative register information from Statistics Denmark from 19992016. Note: ***: p<0.01, **: p<0.05, *: p<0.1. Coefficient estimates from linear probability models. Standard errors clustered by municipality of assignment (94) in parentheses. 't' refers to the year of arrival (asylum). The sample is the subsample of refugee household heads who got asylum during 1999-2010, who were observed in the first four years since arrival and who arrived after

29

the first 10 municipalities had been filled in their calendar year of arrival as well as jointly arrived spouses. Controls (measrued at arrival): Indicators for country of origin, year of arrival, month of arrival, and missing information on educational attainment. Mean of dependent variable: 0.17, i.e. 17% of individuals in the sample have moved out of the assigned municipality within four years since arrival.

30

Table 3. Summary statistics of assigned municipality characteristics at arrival across samples of refugees. Mean (standard deviation).

Balanced panel

Unemployment rate (%) Unemployment rate of non-Western immigrants (%) Employment rate Employment rate of non-Western immigrants (%) Employment growth (%) Population share (%) Non-Western immigrant share (%) Commuting distance to center of local labour market by public transportation Commuting distance to center of local labour market by car Distance to center of local labour market Co-national share (%) Annual influx of assigned refugees per 1,000 inhabitants Number of individuals Number of municipalities

Subsample of balanced panel

Individuals in balanced panel who are excluded from the subsample

Date of asylum during the calendar year of asylum After first 10 Before first 10 municipal Any date municipal quotas quotas were filled were filled (1) (2) (3) 4.1260 4.1266 4.1253 (1.5868) (1.6784) (1.4878) 13.9772 14.5838 13.3584

t-test of difference in mean between (2) and (3)

(4) 0.00132 (0.04) 1.225***

(7.4721)

(7.8798)

(6.9786)

76.6560 (3.8668) 46.3156

76.8125 (3.7759) 46.0402

76.4963 (3.9514) 46.5965

(8.6950)

(9.1746)

(8.1682)

-0.1139 (1.8378) 1.1755 (1.0430) 2.4180 (1.2306) 22.5304 (25.9507)

0.3422 (1.3785) 1.127 (0.9206) 2.2585 (1.1523) 23.5481 (27.149)

-0.5793 (2.1111) 1.2250 (1.1526) 2.5808 (1.2855) 21.4920 (24.6286)

(-2.95) 0.921*** (23.84) -0.0981*** (-4.33) -0.322*** (-12.16) 2.056*** (3.65)

18.0767

18.7411

17.3988

1.342***

(18.2339)

(18.591)

(17.8394)

(3.39)

15.2672 (15.4263) 0.1679 (0.2129) 0.6568

15.8048 (15.665) 0.1440 (0.1925) 0.7339

14.7188 (15.1616) 0.1923 (0.2294) 0.5782

1.086** (3.24) -0.0483*** (-10.51) 0.156***

(0.4462)

(0.4983)

(0.3698)

(16.31)

8,479

4,282

4,197

8,479

95

94

94

(7.57) 0.316*** (3.77) -0.556**

Source: Administrative register information from Statistics Denmark from 1999-2016.

31

Note: Column 1: The balanced panel of household heads comprises the newly arrived adult refugees in the period 1999 to 2010, who were observed for at least four years since arrival (asylum), were in their working ages (aged 18 to 59) at the time of arrival and are household heads. Column 2: The subsample of the balanced panel of household heads is constituted by individuals in the balanced panel of household heads who arrived after the first 10 municipalities had been filled in their calendar year of arrival. Column 3: The subsample of the balanced panel of household heads is constituted by individuals in the balanced panel of household heads who arrived before the first 10 municipalities had been filled in their calendar year of arrival. All municipality characteristics refer to the year of arrival.

32

Table 4. Effects of employment rate in the municipality of residence on individual employment status Dependent variable: Current employment status (year t+s) Years since arrival (s) 2 3 4 Panel A: Balanced panel Current employment rate (year t+s) R2 Mean of dependent variable Number of observations Panel B: Estimation sample Current employment rate (year t+s) R2 Number of observations Panel C: Estimation sample Employment rate at arrival (year t) R2 Mean of dependent variable Number of observations

0.00786*** (0.00168) 0.180 0.2910

0.00641*** (0.00166) 0.209

0.00534*** (0.00175) 0.203 0.2337

0.00728*** (0.00203) 0.155 0.3706 8,479 0.00705*** (0.00161) 0.189 5,096 0.00646*** (0.00163) 0.183 0.3232 5,096

0.00837*** (0.00204) 0.148 0.4094

0.00705*** (0.00258) 0.163

0.00538** (0.00247) 0.162 0.3656

Source: Administrative register information from Statistics Denmark from 1999-2016.

Note: ***: p<0.01, **: p<0.05, *: p<0.1. Coefficient estimates from linear probability models. Standard errors clustered by municipality of assignment (94) in parentheses. 't' refers to the year of arrival (asylum). Panel A: Results from estimation of Eq. 2. The balanced panel of refugee household heads comprises the newly arrived adult refugees in the period 1999 to 2010, who were observed for at least four years since arrival, were in their working ages (aged 18 to 59) at the time of arrival and are household heads. Panel B: Results from estimation of Eq. 2. The estimation sample is constituted by i) individuals who got asylum during 1999-2010, who were observed in the first four years since arrival and who arrived after the first 10 municipalities had been filled in their calendar year of arrival and ii) their spouse, if he/she arrived at the same date. Panel C: Results from estimation of Eq. 1. Same sample as in Panel B. Controls (measured at arrival): Age, indicators for male, marital status, number of children aged 0-2 and 3-17, educational attainment, country of origin, year of arrival, month of arrival as well as assigned municipality characteristics (population share and non-Western immigrants share).

33

Table 5. Robustness checks: Effects of alternative measures of local labour demand at arrival on individual employment status. Dependent variable: Current employment status (year t+s) 2 Panel A: Employment rate at arrival (year t) R2 Panel B: Unemployment rate at arrival (year t) R2 Panel C: Employment rate of non-Western immigrants at arrival (year t) R2 Panel D: Unemployment rate of non-Western immigrants at arrival (year t) R2 Panel E: Employment growth at arrival (year t) R2 Number of observations

Years since arrival (s) 3

4

0.00534*** (0.00175)

0.00646*** (0.00163)

0.00538** (0.00247)

0.203

0.183

0.162

-0.00935* (0.00522)

-0.0140** (0.00583)

-0.0173*** (0.00582)

0.202

0.183

0.163

0.00232*** (0.000778)

0.00224** (0.000955)

0.00152 (0.00123)

0.202

0.182

0.161

-0.00162** (0.000729)

-0.00248*** (0.000876)

-0.00253*** (0.000957)

0.201

0.182

0.162

-0.00392 (0.00901)

-0.00796 (0.00980)

-0.000410 (0.0110)

0.201

0.181 5,096

0.161

Source: Administrative register information from Statistics Denmark from 1999-2016. Note: ***: p<0.01, **: p<0.05, *: p<0.1. Coefficient estimates from linear probability models. Results from estimation of Eq. 1. Standard errors clustered by municipality of assignment (94) in parentheses. 't' refers to the year of arrival (asylum). The sample is the subsample of refugee household heads who got asylum during 1999-2010, who were observed in the first four years since asylum and who arrived after the first 10 municipalities had been filled in their calendar of arrival as well as jointly arrived spouses. Controls (measured at arrival): Age, indicators for male, marital status, number of children aged 0-2 and 3-17, educational attainment, country of origin, year of arrival, month of arrival as well as assigned municipality characteristics (population share and non-Western immigrants share).

34

Table 6. Robustness checks: Effects of local labour demand at arrival on individual employment status. Additional municipality controls. Dependent variable: Employment status four years since arrival (1) (2) (3) (4) (5) (6) Panel A: 0.00538** 0.00542** 0.00550** 0.00547** 0.00523** 0.00415 Employment rate at arrival (year t) (0.00247) (0.00246) (0.00244) (0.00244) (0.00247) (0.00337) R2 Panel B: Unemployment rate at arrival (year t)

0.162

0.162

-0.0173*** -0.0178*** (0.00582) (0.00580)

R2 0.163 0.163 Additional controls relative to the controls in Table 4:

0.162

0.162

-0.0178*** -0.0177*** (0.00575) (0.00576)

0.163

0.171

-0.0171*** (0.00578)

-0.0103 (0.0101)

0.163

0.163

0.164

0.171

No

No

No

Commuting time using public transportation

No

Yes

No

Commuting time by car

No

No

Yes

No

No

No

Commuting distance

No

No

No

Yes

No

No

Co-national share

No

No

No

No

Yes

No

Commuting area F.E. Number of observations

No

No

No

No 5,096

No

Yes

Source: Administrative register information from Statistics Denmark from 1999-2016. Note: ***: p<0.01, **: p<0.05, *: p<0.1. Coefficient estimates from linear probability models. Results from estimation of Eq. 1. Standard errors clustered by municipality of assignment (94) in parentheses. 't' refers to the year of arrival (asylum). The sample is the subsample of refugee household heads who got asylum during 1999-2010, who were observed in the first four years since asylum and who arrived after the first 10 municipalities had been filled in their calendar year of arrival as well as jointly arrived spouses. Additional controls (measured at arrival): Age, indicators for male, marital status, number of children aged 0-2 and 3-17, educational attainment, country of origin, year of arrival, month of arrival and years since arrival, as well as assigned municipality characteristics (population share and non-Western immigrants share).

35

Table 7. Robustness checks: Effects of local labour demand for low-skilled on individual employment status. Dependent variable: Current employment status (year t+s) Years since arrival (s) 2 3 4 Panel A: 0.00524** 0.00652** * * 0.00429** Employment rate of low-skilled at arrival (year t) (0.00146) (0.00141) (0.00213) R2 Panel B: Unemployment rate of low-skilled at arrival (year t) R2 Number of observations

0.203

0.184

0.162

-0.00449 (0.00354)

-0.00847** (0.00379)

0.0103*** (0.00373)

0.201

0.182 5,096

0.163

Source: Administrative register information from Statistics Denmark from 19992016. Note: ***: p<0.01, **: p<0.05, *: p<0.1. Coefficient estimates from linear probability models. Results from estimation of Eq. 1. Standard errors clustered by municipality of assignment (94) in parentheses. 't' refers to the year of arrival (asylum). The sample is the subsample of refugee household heads who got asylum during 1999-2010, who were observed in the first four years since arrival and who arrived after the first 10 municipalities had been filled in their calendar year of arrival as well as jointly arrived spouses. Controls (measured at arrival): Age, indicators for male, marital status, number of children aged 0-2 and 3-17, educational attainment, country of origin, year of asylum, month of asylum and years since asylum, as well as assigned municipality characteristics (population share and non-Western immigrants share).

36

Panel B: Unemployment rate at arrival (year t) Unemployment rate at arrival (year t)*Woman R2 Panel C: Employment rate at arrival (year t) Employment rate at arrival (year t)*More than 12 years of education R2 Panel D: Unemployment rate at arrival (year t) Unemployment rate at arrival (year t)*More than 12 years of education R2 Number of observations

-0.00936 (0.00641) 3.37e-05 (0.00674)

-0.0125 (0.00762) -0.00523 (0.00858)

-0.0185*** (0.00666) 0.00427 (0.00827)

0.202

0.183

0.163

0.00526*** (0.00190) 0.000530 (0.00341)

0.00635*** (0.00186) 0.000700 (0.00463)

0.00625** (0.00251) -0.00566 (0.00529)

0.203

0.183

0.162

-0.00786 (0.00589) -0.00973 (0.00888)

-0.0124* (0.00667) -0.0106 (0.0120)

-0.0185*** (0.00595) 0.00831 (0.0115)

0.202

0.183 5,096

0.163

Source: Administrative register information from Statistics Denmark from 1999-2016. Note: ***: p<0.01, **: p<0.05, *: p<0.1. Coefficient estimates from linear probability models. Results from estimation of Eq. 1. Standard errors clustered by municipality of assignment (94) in parentheses. 't' refers to the year of arrival (asylum). The sample is the subsample of refugee household heads who got asylum during 1999-2010, who were observed in the first four years since arrival and who arrived after the first 10 municipalities had been filled in their year of arrival as well as jointly arrived spouses. Controls (measured at arrival): Age, indicators for male, marital status, number of children aged 0-2 and 3-17, educational attainment, country of origin, year of asylum, month of asylum and years since asylum, as well as assigned municipality characteristics (population share and non-Western immigrants share). Share of individuals with more than 12 years of education: 14.40%.

37

Fig. 1 Figure 1. Geographical stability of settlement for refugees arriving 1999-2010

Source : Administrative register information from Statistics Denmark 1999-2016. Note : Calculation is based on a gross sample of refugees including all adult refugees arriving to Denmark from 1999-2010. The out-migration rate from the municipality of assignment is calculated as Kaplan-Meier empirical hazard rates. The rate of staying in the municipality of assignment is calculated as Kaplan-Meier empirical survivor rates.

38

Figure 2. Cumulative distribution function for municipalities that have met their annual quota of refugees by a given calendar month. Separate figures for each calendar year, 1999-2010.

Source : Administrative register information from Statistics Denmark from 1999-2010. Note : The sample is a gross sample of refugee household heads who got asylum during 1999-2010.

39

Figure A1. Correlation between predicted annual refugee quota and actual annual refugee quotas across municipalities in 2010

Source: Administrative register information from Statistics Denmark from 2009-2010 and https://www.nyidanmark.dk/da/Numbers/visiteringskvoter. Note: The correlation between actual and calculated quotas in the 98 muncipalities is 0.947.

40

Table A1. Balancing tests for different subsamples of the balanced panel of household heads Dependent variable: Assigned municipality characteristics at arrival Une mpl oymen t rate

Une mpl oymen t rate of nonWes tern imm igra nts

Em plo yme nt rat e

Em ploy men t rate of nonWes tern imm igra nts

Em plo yme nt gro wth

Pop ulati on shar e

Non Wes tern imm igra nts shar e

Com muti ng dista nce to cent er of local labo ur mar ket by publ ic trans portatio n

Com muti ng dista nce to cent er of local labo ur mar ket by car

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

0.00 350

0.0 237 (0. 025 3) 0.0 026 8 (0. 030 1) 0.8 09

0.02 80

1.19 4

0.89 6

(0.0 365)

(0.8 13)

(0.5 76)

0.01 0.06 0.63 More than 12 years 42 44 3 (0.0 (0.0 (0.2 (0.2 350 (0.0 (0.9 513) 58) 78) ) 406) 16) 0.29 0.24 0.33 0.04 0.13 0.05 R2 9 7 8 6 4 0 Number of municipalities 95 Number of individuals 8,479 Panel B: Subsample of balanced panel of HH who arrived after first 4 municipalities had filled their annual refugee quota Educational attainment (reference category: 0-9 years) at arrival: 0.0 0.03 0.04 93 0.06 0.0 0.01 0.02 0.91 10-12 years 18 36 6 33 416 22 31 1 (0. (0. (0.0 (0.0 (0.2 12 (0.2 026 337 (0.0 (0.8 496) 51) 5) 60) 7) ) 387) 65) 0.3 0.0 0.09 0.28 31 0.30 004 0.01 0.05 0.19 More than 12 years 89* 6 ** 7 56 23 40 5 (0. (0. (0.0 (0.0 (0.2 14 (0.2 031 362 (0.0 (0.9 550) 76) 4) 98) 8) ) 427) 93)

0.86 8

Panel A: Balanced panel of household heads (HH) Educational attainment (reference category: 0-9 years) at arrival: 0.01 0.02 10-12 years 52 07 (0.0 459)

(0.2 33)

0.07 06

0.16 7

0.0 41 3 (0. 11 6) 0.2 47 * (0. 13 5) 0.1 47

(0.2 41)

0.24 5

0.02 25 (0.0 319 )

(0.6 45) 0.04 9

0.75 8 (0.6 18) 0.57 3 (0.6 95)

Dis tan ce to cen ter of loc al lab or ma rke t

Conati ona l sha re

(10 )

(11 )

0.8 04 * (0. 48 8)

0.0 056 (0.0 049 4)

0.5 69 (0. 54 5) 0.0 48

0.0 015 (0.0 053 7) 0.4 36

0.7 15 (0. 52 2) 0.2 58 (0. 58 6)

0.0 042 (0.0 051 9) 0.0 081 (0.0 057 6)

41

0.29 0.24 0.1 0.33 0.7 0.04 0.12 0.05 R2 4 5 32 3 94 6 7 2 Number of municipalities 95 Number of individuals 7,409 Panel C: Subsample of balanced panel of HH who arrived after first 8 municipalities had filled their annual refugee quota Educational attainment (reference category: 0-9 years) at arrival: 0.01 0.01 0.1 0.01 0.0 0.01 0.01 0.87 10-12 years 13 01 03 92 323 58 57 9 (0. (0. (0.0 (0.0 (0.2 13 (0.2 029 371 (0.0 (0.9 552) 85) 8) 89) 9) ) 425) 58) 0.3 0.07 0.33 14 0.28 0.0 0.01 0.05 0.14 More than 12 years 81 6 ** 4 301 91 24 7 (0. (0. (0.0 (0.0 (0.3 15 (0.3 034 378 (0.0 (1.1 611) 06) 9) 31) 9) ) 457) 01) 0.29 0.24 0.1 0.33 0.7 0.04 0.12 0.44 2 R 4 5 32 3 94 6 7 2 Number of municipalities 95 Number of individuals 5,952 Panel D: Subsample of balanced panel of HH who arrived after first 9 municipalities had filled their annual refugee quota Educational attainment (reference category: 0-9 years) at arrival: 0.01 0.10 0.1 0.10 0.0 0.02 0.03 0.95 10-12 years 35 2 44 4 238 07 54 1 (0. (0. (0.0 (0.0 (0.2 14 (0.3 030 377 (0.0 (1.0 587) 99) 7) 12) 7) ) 442) 27) 0.2 0.07 0.18 86 0.16 0.0 0.03 0.05 0.20 More than 12 years 24 4 * 3 555 04 71 0 (0. (0. (0.0 (0.3 17 (0.3 034 (0.0 (0.0 (1.2 662) 34) 1) 66) 9) 35) 452) 07) 0.31 0.26 0.1 0.33 0.7 0.05 0.12 0.07 R2 3 9 18 2 18 0 7 0 Number of municipalities 95 Number of individuals 5,185 Panel E: Subsample of balanced panel of HH who arrived after first 10 municipalities had filled their annual refugee quota Educational attainment (ref. category 0-9 years) at arrival: 0.00 0.13 0.1 0.11 0.0 0.01 0.04 0.75 10-12 years 21 0 30 4 152 49 00 0 (0. (0. (0.0 (0.0 (0.3 16 (0.3 033 402 (0.0 (1.1 644) 25) 0) 43) 1) ) 480) 25) 0.04 0.02 0.2 0.12 0.0 0.03 0.07 0.96 More than 12 years 29 49 31 9 521 42 70 7 (0. (0. (0.0 (0.0 (0.3 18 (0.4 037 365 (0.0 (1.3 731) 71) 6) 07) 4) ) 490) 32)

0.04 9

0.0 48

0.4 42

0.87 9

0.7 81 (0. 68 2) 0.3 06 (0. 76 3) 0.0 49

0.0 057 (0.0 056 2) 0.0 027 (0.0 059 9) 0.0 48

(0.9 58) 0.14 7 (1.1 01) 0.05 2

0.89 1 (0.7 27) 0.52 5 (0.8 30) 0.06 5

0.59 3 (0.7 93) 0.90 7 (0.9 17)

0.8 88 (0. 61 3) 0.2 46 (0. 69 6) 0.0 65

0.5 94 (0. 66 9) 0.6 41 (0. 76 9)

0.0 052 (0.0 059 4) 0.0 020 (0.0 064 8) 0.4 40

0.0 055 (0.0 060 3) 0.0 054 (0.0 069 3)

42

0.34 0.28 R2 3 6 Number of municipalities Number of individuals Panel F: Subsample of balanced panel of HH who arrived after first 11 municipalities had filled their annual refugee quota Educational attainment (reference category: 0-9 years) at arrival: 0.01 0.03 10-12 years 46 32

More than 12 years

(0.0 712) 0.06 93

(0.3 60) 0.25 9

(0.0 788) 0.37 5

(0.3 99) 0.31 0

R2 Number of municipalities Number of individuals Panel G: Subsample of balanced panel of HH who arrived after first 14 municipalities had filled their annual refugee quota Educational attainment (reference category: 0-9 years) at arrival: 0.00 0.00 10-12 years 362 726

More than 12 years

R2 Number of municipalities Number of individuals

(0.0 915) 0.04 22

(0.4 67) 0.21 3

(0.1 01) 0.39 2

(0.5 05) 0.35 9

0.1 22

0.1 30 (0. 17 9) 0.2 10 (0. 20 4) 0.1 23

0.1 23 (0. 22 4) 0.2 85 (0. 25 8) 0.1 22

0.35 1

0.7 12

0.05 0.13 5 6 94 4,282

0.02 53

0.0 134 (0. 035 9) 0.0 614 (0. 040 8) 0.7 33

0.03 74 (0.0 446 ) 0.02 03 (0.0 409 ) 0.06 7 94 3,627

0.0 110 (0. 044 8) 0.0 588 (0. 048 1) 0.7 41

0.00 38 (0.0 577 )

(0.3 82) 0.19 5 (0.4 48) 0.37 5

0.07 77 (0.4 72) 0.57 0 (0.5 52) 0.43 8

0.01 11 (0.0 501 ) 0.09 1 93 2,499

0.07 5

0.06 7

0.0 67

0.05 24

0.83 6

0.45 2

(0.0 531) 0.14 2** *

(1.2 74) 0.94 0

(0.8 84) 1.22 4

(0.0 547) 0.15 5

(1.5 01) 0.09 0

(1.0 20) 0.08 0

0.5 22 (0. 74 6) 0.9 12 (0. 85 4) 0.0 81

0.05 62

0.36 2

0.11 3

0.16 9**

(1.6 69) 2.08 6

(1.1 27) 1.79 8

(0.0 679) 0.18 4

(1.8 80) 0.07 5

(1.2 58) 0.07 8

(0.0 673)

0.1 30 (0. 94 2) 1.3 21 (1. 04 9) 0.0 80

0.4 30

0.0 047 (0.0 067 7) 0.0 120 (0.0 077 5) 0.4 39

0.0 003 (0.0 083 8) 0.0 140 (0.0 091 1) 0.4 49

Source: Administrative register information from Statistics Denmark from 1999-2016. Note: ***: p<0.01, **: p<0.05, *: p<0.1. Coefficient estimates from linear regression. Standard errors clustered by municipality of assignment in parentheses. Panel A: The balanced panel of refugee household heads comprises the newly arrived adult refugees in the period 1999 to 2010, who were observed for at least four years since arrival (asylum), were in their working ages (aged 18 to 59) at the time of arrival and are household heads. Panels B-G: Subsamples of the balanced panel of household heads. Additional controls (measured at arrival): Age, indicators for male, marital status, number of children aged 0-2 and 3-17, country of origin, year of asylum, month of asylum as well as assigned municipalitycharacteristics (population share and non-Western immigrants share).

43

Table A2. Alternative balancing tests: Educational attainment of assignees (household heads) and assigned municipality characteristics at arrival. Dependent variable: Household head has at least 10 years of education at arrival (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Assigned municipality characteristics at arrival 0.00 Unemployment rate 151 (0.0 047 5) 0.00 024 8 (0.0 Unemployment rate among 010 Non-Western immigrants 1) 0.00 Employment rate 211 (0.0 017 7) Employment rate among Non0.00 Western immigrants 034 4 (0.0 008 82) 0.00 Employment growth 308 (0.0 082 8) 0.00 Population share 437 (0.0 065 2) 0.00 Non-Western immigrant share 740 (0.00 555)

Commuting time to center of local labour market by public transportation

Commuting time to center of local labour market by car

2.00 e-05 (0.0 002 33) 6.74 e-06 (0.0 003 44

42) 8.01 e-05 (0.0 004 05)

Distance to center of local labour market

0.00 967 (0.04 00)

Co-national share

0.02 57 (0.0 159)

Annual influx of assigned refugees Controls:

Man

Age

Married

Children aged 0-2

Children aged 3-17

Country of origin F.E. Year of arrival F.E. Month of arrival F.E. R2 Number of municipalities Number of observations

0.09 42* ** (0.0 171) 0.00 724 *** (0.0 007 74) 0.01 06 (0.0 168) 0.02 79 (0.0 193) 0.01 47 (0.0 164) Yes Yes Yes 0.32 6

0.09 38* ** (0.0 171) 0.00 724 *** (0.0 007 73) 0.01 06 (0.0 168) 0.02 79 (0.0 193) 0.01 47 (0.0 164) Yes Yes Yes 0.32 6

0.09 51* ** (0.0 172) 0.00 721 *** (0.0 007 74) 0.01 02 (0.0 168) 0.02 76 (0.0 193) 0.01 48 (0.0 164) Yes Yes Yes 0.32 7

0.09 42* ** (0.0 171) 0.00 723 *** (0.0 007 74) 0.01 05 (0.0 168) 0.02 79 (0.0 193) 0.01 46 (0.0 164) Yes Yes Yes 0.32 6

0.09 40* ** (0.0 171) 0.00 724 *** (0.0 007 74) 0.01 08 (0.0 168) 0.02 79 (0.0 193) 0.01 47 (0.0 164) Yes Yes Yes 0.32 6

0.09 43* ** (0.0 171) 0.00 723 *** (0.0 007 74) 0.01 07 (0.0 168) 0.02 80 (0.0 193) 0.01 44 (0.0 164) Yes Yes Yes 0.32 6

0.09 40* ** (0.0 171) 0.00 726 *** (0.0 007 74) 0.01 00 (0.0 168) 0.02 82 (0.0 193) 0.01 52 (0.0 164) Yes Yes Yes 0.32 7

0.09 40* ** (0.0 171) 0.00 724 *** (0.0 007 74) 0.01 06 (0.0 168) 0.02 79 (0.0 193) 0.01 46 (0.0 164) Yes Yes Yes 0.32 6

0.09 40* ** (0.0 171) 0.00 724 *** (0.0 007 74) 0.01 06 (0.0 168) 0.02 79 (0.0 193) 0.01 47 (0.0 164) Yes Yes Yes 0.32 6

0.09 41* ** (0.0 171) 0.00 724 *** (0.0 007 74) 0.01 06 (0.0 168) 0.02 80 (0.0 193) 0.01 46 (0.0 164) Yes Yes Yes 0.32 6

0.09 40* ** (0.0 171) 0.00 724 *** (0.0 007 74) 0.01 07 (0.0 168) 0.02 80 (0.0 193) 0.01 45 (0.0 164) Yes Yes Yes 0.32 6

0.09 40* ** (0.0 171) 0.00 728 *** (0.0 007 73) 0.01 11 (0.0 168) 0.02 83 (0.0 193) 0.01 56 (0.0 164) Yes Yes Yes 0.32 7

94 4,282

Source: Administrative register information from Statistics Denmark from 1999-2016. Note: ***: p<0.01, **: p<0.05, *: p<0.1. Robust standard errors in parentheses. The sample is the subsample of refugee household heads who got asylum during 1999-2010, who were observed in the first four years since arrival (asylum) and who arrived after the

45

first 10 municipalities had been filled in their year of arrival. Reported coefficients are based on linear regressions of an indicator for having at least ten years of education at arrival on an assigned municipality characteristic and other individual characteristics in the year of arrival. Indicator for missing information on educational attainment included.

Table A3. Effects of assigned local unemployment rate on the individual's probability of relocation Dependent variable: Moved out of the assigned municipality within four years since arrival (1) (2) (3) (4) (5) (6) Individual characteristics at arrival Educational attainment (reference category: 0-9 years) 10-12 years 0.00112 (0.0143) More than 12 years 0.0148 (0.0172) Male 0.0375*** (0.0101) 0.00394** Age * (0.000698 ) Married

-0.0393** (0.0152) Children aged 0-2 0.0465*** (0.0146) Children aged 3-17 -0.106*** (0.0139) Characteristics of assigned municipality at arrival Unemployment rate at arrival (year t) 0.0155*** (0.00444)

0.000745 (0.0142) 0.0150 (0.0172) 0.0373*** (0.0101) 0.00394** * (0.000697 ) 0.0398*** (0.0151) 0.0457*** (0.0146) -0.107*** (0.0139)

0.000493 (0.0142) 0.0147 (0.0171) 0.0373*** (0.0101) 0.00393** * (0.000697 ) 0.0400*** (0.0152) 0.0457*** (0.0146) -0.107*** (0.0139)

0.000362 (0.0142) 0.0150 (0.0172) 0.0373*** (0.0101) 0.00395** * (0.000699 )

0.0157*** (0.00443)

-0.00747 (0.01084) 0.03978** * (0.00904) -0.0581 (0.0470)

Unemployment rate of non-Western immigrants at arrival (year t) Population share at arrival (year t)

Non-Western immigrant share at arrival (year t) Co-national share at arrival (year t) Commuting time using public transportation Commuting time by car

-0.00902 (0.01078) 0.04202** * (0.00899)

-0.0397** (0.0152) 0.0458*** (0.0146) -0.107*** (0.0139)

0.000500 (0.0143) 0.0147 (0.0172) 0.0373*** (0.0101) 0.00393** * (0.000700 ) 0.0400*** (0.0152) 0.0457*** (0.0146) -0.107*** (0.0139)

0.000483 (0.0143) 0.0147 (0.0172) 0.0373*** (0.0101) 0.00393** * (0.000701 ) 0.0400*** (0.0152) 0.0457*** (0.0146) -0.107*** (0.0139)

0.0118 (0.00732)

0.0118 (0.00724)

0.0118 (0.00739)

0.0118 (0.00739)

0.00118 (0.00165) -0.00718 (0.01107) 0.04049** * (0.00914) -0.0586 (0.0471)

0.00129 (0.00167) -0.00552 (0.01134) 0.04015** * (0.909) -0.0586 (0.0469) 0.000188 (0.000282 )

0.00118 (0.00165) -0.00728 (0.01214) 0.04050** * (0.00915) -0.0587 (0.0468)

0.00118 (0.00165) -0.00708 (0.01207) 0.04048** * (0.00916) -0.0585 (0.0466)

-1.45e-05 (0.000473 46

) Commuting distance

1.83e-05 (0.000570 )

R2 Number of observations

0.118

0.118

0.118

0.119

0.118

0.118

5,096

Source: Administrative register information from Statistics Denmark from 1999-2016. Note: ***: p<0.01, **: p<0.05, *: p<0.1. Coefficient estimates from linear probability models. Mean of dependent variable: 0.17, i.e. 17% of individuals in the sample have moved out of the assigned municipality within four years since asylum. Standard errors clustered by municipality of assignment (94) in parentheses. 't' refers to the year of arrival (asylum). The sample is the subsample of refugee household heads who got asylum during 1999-2010, who were observed in the first four years since asylum and who arrived after the first 10 municipalities had been filled in their calendar year of arrival as well as jointly arrived spouses. Controls (measured at arrival): Indicators for country of origin, year of arrival, month of arrival, and missing information on educational attainment.

Table A4. Time persistency of local labour market conditions Year t+1

t+2

t+3

t+4

Panel A: t=1999-2014

Dependent variable: Current employment rate (year t+s)

Employment rate in year t

0.971*** (0.00678)

R2 Number of observations Panel B: t=1999-2014 Unemployment rate in year t

0.908*** (0.0110)

0.863*** (0.0141)

0.851*** (0.0160)

0.929 0.814 0.717 0.674 1,568 1,568 1,470 1,372 Dependent variable: Current unemployment rate (year t+s) 0.789***

0.553***

0.380***

0.276***

(0.0146)

(0.0199)

(0.0228)

(0.0247)

R2 Number of observations Panel C: t=1999-2002

0.667 0.343 0.166 0.095 1,568 1,568 1,470 1,372 Dependent variable: Current employment rate (year t+s)

Employment rate in year t

0.978*** (0.00931)

R2 Number of observations Panel D: t=2003-2010 Employment rate in year t

0.966 0.907 0.901 0.950 392 392 392 392 Dependent variable: Current employment rate (year t+s) 0.961*** 0.870*** 0.815*** 0.801*** (0.0133) (0.0216) (0.0263) (0.0278)

R2 Number of observations

0.870 784

0.956*** (0.0155)

0.675 784

0.950*** (0.0160)

0.551 784

0.960*** (0.0111)

0.515 784

Source: Administrative registers from Statistics Denmark for the period 1999-2014. Note: *** p<0.01, ** p<0.05, * p<0.1. Coefficient estimates from OLS regressions. Standard errors in parentheses. 't' refers to the year of asylum. The table includes information from 98 municipalities over a period of 16 years.

47

Table A5. Effects of assigned municipality characteristics on individual employment status

Employment rate at arrival (year t) Population share at arrival (year t) Non-Western immigrants share at arrival (year t)

Dependent variable: Current employment status (year t+s) Years since arrival (s) 3 2 4 0.00534*** 0.00646*** 0.00538** (0.00175) (0.00163) (0.00247) 0.00627 0.0179* 0.0196* (0.00786) (0.0105) (0.0109) 0.00729 -0.00118 -0.0142* (0.00674) (0.00736) (0.00801)

R2

0.203

Number of observations

0.183 5,096

0.162

Source: Administrative register information from Statistics Denmark from 19992016. Note: ***: p<0.01, **: p<0.05, *: p<0.1. Coefficient estimates from linear probability models. Results from estimation of Eq. 1. Standard errors clustered by municipality of assignment (94) in parentheses. 't' refers to the year of arrival (asylum). The sample is the subsample of refugee household heads who got asylum during 1999-2010, who were observed in the first four years since arrival and who arrived after the first 10 municipalities had been filled in their calendar year of arrival as well as jointly arrived spouses. Additional controls (measured at arrial): Age, indicators for male, marital status, number of children aged 0-2 and 3-17, educational attainment, country of origin, year of arrival, month of arrival and years since arrival, as well as municipality of assignment characteristics (population share and non-Western immigrants share).

Table A6. Effects of assigned employment rate on individual employment status. Selected subsamples. Dependent variable: Current employment status (year t+s) Years since asylum (s) 2 3 4 Panel A: Balanced panel of refugee household heads Employment rate at arrival (year t)

0.00607***

0.00641***

0.00659***

(0.00165)

(0.00197)

(0.00200)

R2 0.173 0.152 0.146 Mean of dependent variable 0.2911 0.3706 0.4157 Number of assigned municipalities 95 Number of observations 8,479 Panel B: Subsample of balanced panel of HH who arrived after first 4 municipalities had filled their annual refugee quota Employment rate at arrival (year t) 0.00573*** 0.00670*** 0.00612*** (0.00159) 2

R Mean of dependent variable Number of assigned municipalities

(0.00189)

(0.00196)

0.178

0.156

0.147

0.2875

0.3704

0.4164

95 48

Number of observations 7,409 Panel C: Subsample of balanced panel of HH who arrived after first 8 municipalities had filled their annual refugee quota Employment rate at arrival (year t) 0.00544*** 0.00626*** 0.00622*** (0.00174)

(0.00177)

(0.00201)

2

R 0.189 0.165 0.154 Mean of dependent variable 0.2762 0.3634 0.4081 Number of assigned municipalities 95 Number of observations 5,952 Panel D: Subsample of balanced panel of HH who arrived after first 9 municipalities had filled their annual refugee quota Employment rate at arrival (year t) 0.00531*** 0.00615*** 0.00561*** (0.00177)

(0.00168)

(0.00212)

2

R 0.195 0.172 0.157 Mean of dependent variable 0.2780 0.3670 0.4089 Number of assigned municipalities 95 Number of observations 5,185 Panel E: Subsample of balanced panel of HH who arrived after first 10 municipalities had filled their annual refugee quota Employment rate at arrival (year t) 0.00556*** 0.00635*** 0.00536** (0.00206)

(0.00186)

(0.00250)

R2 0.199 0.177 0.159 Mean of dependent variable 0.2606 0.3524 0.3895 Number of assigned municipalities 94 Number of observations 4,282 Panel F: Subsample of balanced panel of HH who arrived after first 10 municipalities had filled their annual refugee quota and their jointly arrived spouses (baseline sample) Employment rate at arrival (year t)

0.00534***

0.00646***

0.00538**

(0.00175)

(0.00163)

(0.00247)

2

R 0.203 0.183 0.162 Mean of dependent variable 0.2337 0.3232 0.3656 Number of assigned municipalities 94 Number of observations 5,096 Panel G: Subsample of balanced panel of HH who arrived after first 11 municipalities had filled their annual refugee quota Employment rate at arrival (year t) 0.00487** 0.00470** 0.00573** (0.00203)

(0.00192)

(0.00244)

R2 0.209 0.182 0.164 Mean of dependent variable 0.2633 0.3535 0.3921 Number of assigned municipalities 94 Number of observations 3,627 Panel H: Subsample of balanced panel of HH who arrived after first 14 municipalities had filled their annual refugee quota Employment rate at arrival (year t) 0.00234 0.00422* 0.00508* R

2

(0.00217)

(0.00226)

(0.00301)

0.216

0.189

0.179

49

Mean of dependent variable Number of assigned municipalities Number of observations

0.2589

0.3413

0.3894

93

2,499 Source: Administrative register information from Statistics Denmark from 1999-2016. Note: ***: p<0.01, **: p<0.05, *: p<0.1. Coefficient estimates from linear probability models. Results from estimation of Eq. 1. Standard errors clustered by municipality of assignment in parentheses. 't' refers to the year of arrival (asylum). Panel A: The balanced panel of refugee household heads comprises the newly arrived adult refugees in the period 1999 to 2010, who were observed for at least four years since arrival, were in their working ages (aged 18 to 59) at the time of arrival and are household heads. Panels B-H: Subsamples of the balanced panel of household heads. Controls (measured at arrival): Age, indicators for male, marital status, number of children aged 0-2 and 3-17, educational attainment, country of origin, year of arrival, month of arrival as well as municipality of assignment characteristics (population share and non-Western immigrants share).

Table A7. Effects of assigned employment rate on individual employment status by immigration cohort groups Dependent variable: Current employment status (year t+s) Years since arrival (s) 2 3 4 Panel A: Immigration cohorts 1999-2002 Employment rate at arrival (year t) R2 Number of observations Panel B: Immigration cohorts 2003-2010 Employment rate at arrival (year t)

0.00650*** (0.00195)

0.00958*** (0.00199)

0.00785** (0.00323)

0.175 3,335

0.164 3,335

0.159 3,335

0.00184 (0.00433)

-0.00126 (0.00363)

-0.000229 (0.00359)

0.262 1,761

0.246 1,761

0.197 1,761

Yes

Yes

Yes

Yes Yes Yes Yes

Yes Yes Yes Yes

Yes Yes Yes Yes

Yes

Yes

Yes

R2 Number of observations Controls: Age and dummies for male, marital status, number of children aged 0-2, number of children aged 3-17 Educational attainment F.E. Country of origin F.E. Year of arrival F.E. Month of arrival F.E. Assigned municipality characteristics: Population share, non-Western immigrants share Source: Administrative register information from Statistics Denmark from 1999-2016.

Note: ***: p<0.01, **: p<0.05, *: p<0.1. Coefficient estimates from linear probability models. Results from estimation of Eq. 1. Standard errors clustered by municipality of assignment (94) in parentheses. 't' refers to the year of arrival (asylum). The sample is the subsample of refugee household heads who got asylum during 1999-2010, who were observed in the first four years since arrival and who arrived after the first 10 municipalities had been filled in their calendar year of arrival as well as jointly arrived spouses.

Table B1. Sample selection criteria. Number of individuals. 50

#

Explanation 1 All the population in the residence permit register (OPHG) for the period 1997-2016 Sample size after appending ophg files from 1997 to 2016 2 Drop all the observations without pnr (If pnr==.) 3 Drop all the individuals imputed residence permit type in any given year If an individual’s observation is imputed at any time 4 Keep only 1st residence permit of each individual 5 Keep only the refugees 6 Residence permit between 1999 and 2010 7 Country of origin of the individual or spouse not Denmark 8 The individual is found in the population register (BEF) at least once between 1999 to 2016 9 Age at arrival is between 18 and 59, calculated as

Reductio n

Sample size Panel

1,223,63 3

242,878

980,755

204,523

776,232

199,642

576,590

520,607

55,983

36,988

18,995

164

18,831

655

18,176

5,764

12,412

51

the year of receiving residence permit minus the date of birth recorded in BEF 10 First appearance in BEF is the same year or one after receiving residence permit , out of which: Observed in year 0 Observed in year 1 11 Household head (below broken down) Without partner (pnrp==.) First refugee with a refugee partner arriving later Man with a refugee partner arriving together 12 Observed in the employment register (RAS) for the period 1999-2015 in years 2-4 Individuals in balanced panel of household heads (HH) Of which received residence permit after 10 municipalities' annual quotas had been filled

197

12,215

63 12,152 3,192

9,023

4,684

2,512

1,827

544

8,479

8,479

4,282

52

(subsample of balanced panel of HH, used in balancing tests) Jointly arrived spouses of HHs in subsample of balanced panel of HH Refugee spouses in working age getting asylum on the same date as the HH who received residence permit once 10 municipalites were filled Of which observed in RAS in years 2-4 Total uncontaminate d subsample (subsample of HH and jointly arrived couples)

1,729

915

814

5,096

Source: Administrative register information from Statistics Denmark from 19992016.

Table B2. Summary statistics of demographic and socioeconomic characteristics of refugees across samples of refugees. Mean (standard deviation). 3) Subsampl 4) Estimation sample e Date of asylum during the calendar year of asylum After first 10 municipal After first 10 municipal Any date Any date quotas quotas were filled were filled Men Women Men Women All Men Women All (7) (8) (1) (2) (3) (4) (5) (6)

1) Gross sample

2) Balanced panel

53

Panel A: Personal attributes Male

Age at registration Married

Having a child aged 0-2

Having a child aged 3-17

1.00

0.00

1.00

0.00

0.82

(0.0000)

(0.0000)

(0.0000)

(0.0000)

(0.3827)

33.49 (10.45) 0.5796

34.55 (12.33) 0.6482

32.59 (90.96) 0.5967

33.18 (10.80) 0.3975

32.66 (9.42) 0.5631

(0.4936)

(0.4776)

(0.4906)

(0.4895)

(0.4961)

0.1068

0.2395

0.1465

0.1031

0.1397

(0.3089)

(0.4268)

(0.3537)

(0.3042)

(0.3467)

0.2294

0.4769

0.3980

0.4119

0.3921

(0.4205)

(0.4995)

(0.4895)

(0.4923)

(0.4883)

0.4466

0.3338

0.4067

0.298

(0.4812)

(0.4972)

(0.4716)

(0.4914)

(0.4574)

0.2813

0.3203

0.2951

0.3460

0.3354

(0.4497)

(0.4667)

(0.4561)

(0.4758)

(0.4722)

0.2064

0.1545

0.2166

0.1540

0.2146

(0.4048)

(0.3614)

(0.4120)

(0.3611)

(0.4106)

0.1481

0.0785

0.1545

0.0933

0.152

(0.3552)

(0.2990)

(0.3614)

(0.2910)

(0.3591)

0.1198

0.1139

0.1237

0.1312

0.2174

(0.3247)

(0.3178)

(0.3292)

(0.3377)

(0.4125)

0.1740

0.1533

0.1760

0.1580

0.1518

(0.3792)

(0.3603)

(0.3809)

(0.3648)

(0.3589)

0.2223

0.1885

0.2215

0.1991

0.2966

(0.4158)

(0.3911)

(0.4153)

(0.3994)

(0.4568)

0.1043

0.1260

0.1006

0.1436

0.0033

(0.3056)

(0.3319)

(0.3008)

(0.3508)

(0.0571)

0.0700

0.0801

0.0695

0.0698

0.068

(0.2552)

(0.2716)

(0.2544)

(0.2550)

(0.2517)

0.0317

0.0545

0.0304

0.0444

0.0241

(0.1751)

(0.2271)

(0.1716)

(0.2060)

(0.1532)

0.0308

0.0398

0.0315

0.0450

0.0381

(0.1729)

(0.1956)

(0.1747)

(0.2075)

(0.1914)

Panel B: Educational attainment Unknown education 0.3642

0-9 years

10-12 years

More than 12 years

Panel C: Asylum year 1999

2000

2001

2002

2003

2004

2005

1.00 (0.0000 ) 32.58 (9.05) 0.6044 (0.4890 ) 0.1469 (0.3541 ) 0.3893 (0.4877 )

0.00 (0.0000 ) 32.47 (9.27) 0.6639 (0.4725 ) 0.2708 (0.4445 ) 0.5910 (0.4918 )

0.69 (0.4623 ) 32.54 (9.12) 0.6228 (0.4847 ) 0.1852 (0.3885 ) 0.4517 (0.4977 )

0.2819 (0.4500 ) 0.3242 (0.4682 ) 0.2299 (0.4208 ) 0.1640 (0.3703 )

0.3519 (0.4777 ) 0.3748 (0.4842 ) 0.1737 (0.3790 ) 0.0996 (0.2995 )

0.3036 (0.4598 ) 0.3399 (0.4737 ) 0.2125 (0.4091 ) 0.1440 (0.3512 )

0.2137 (0.4100 ) 0.1577 (0.3645 ) 0.2987 (0.4577 ) 0.0026 (0.0505 ) 0.0688 (0.2531 ) 0.0224 (0.1482 ) 0.0375 (0.1900 )

0.1934 (0.3951 ) 0.1446 (0.3518 ) 0.2708 (0.4445 ) 0.0051 (0.0711 ) 0.0881 (0.2836 ) 0.0342 (0.1819 ) 0.0399 (0.1959 )

0.2074 (0.4055 ) 0.1536 (0.3606 ) 0.2900 (0.4538 ) 0.0033 (0.0577 ) 0.0748 (0.2630 ) 0.0261 (0.1594 ) 0.0383 (0.1919 ) 54

2006

2007

2008

2009

2010

0.0356

0.0370

0.0373

0.0359

0.0383

(0.1853)

(0.1889)

(0.1895)

(0.1861)

(0.1919)

0.0363

0.0524

0.0337

0.0326

0.0329

(0.1871)

(0.2229)

(0.1804)

(0.1777)

(0.1785)

0.0431

0.0585

0.0430

0.0503

0.0556

(0.2031)

(0.2347)

(0.2030)

(0.2186)

(0.2291)

0.0464

0.0347

0.0465

0.0424

0.0654

(0.2104)

(0.1831)

(0.2106)

(0.2016)

(0.2472)

0.0857

0.0610

0.0862

0.0477

0.0086

(0.2800)

(0.2394)

(0.2807)

(0.2131)

(0.0926)

0.0375 (0.1900 ) 0.0313 (0.1740 ) 0.0546 (0.2272 ) 0.0659 (0.2482 ) 0.0094 (0.0964 )

0.0412 (0.1989 ) 0.0520 (0.2221 ) 0.0736 (0.2611 ) 0.0488 (0.2156 ) 0.0082 (0.0904 )

0.0387 (0.1928 ) 0.0377 (0.1904 ) 0.0604 (0.2383 ) 0.0606 (0.2387 ) 0.0090 (0.0946 )

Table B2. Summary statistics of demographic and socioeconomic characteristics of refugees across samples of refugees. Mean (standard deviation). 3) Subsampl 4) Estimation sample e Date of asylum during the calendar year of asylum: After first 10 municipal After first 10 municipal Any date Any date quotas quotas were filled were filled Men Women Men Women All Men Women All (7) (8) (1) (2) (3) (4) (5) (6)

1) Gross sample

Panel D: Month of asylum January

February

March

April

May

June

July

2) Balanced panel

0.0679

0.0697

0.0662

0.0744

0.0000

(0.2515)

(0.2546)

(0.2487)

(0.2625)

(0.0000)

0.0831

0.0881

0.0832

0.0829

0.0014

(0.2760)

(0.2834)

(0.2762)

(0.2758)

(0.0374)

0.0771

0.0767

0.0759

0.0796

0.0255

(0.2668)

(0.2661)

(0.2648)

(0.2708)

(0.1575)

0.0720

0.0750

0.0720

0.0731

0.0189

(0.2585)

(0.2635)

(0.2585)

(0.2604)

(0.1362)

0.0855

0.0946

0.0838

0.0875

0.064

(0.2796)

(0.2927)

(0.2771)

(0.2826)

(0.2448)

0.1337

0.1466

0.1382

0.1599

0.1448

(0.3403)

(0.3537)

(0.3451)

(0.3667)

(0.3519)

0.1144

0.1100

0.1165

0.1038

0.146

0.0000 (0.0000 ) 0.0011 (0.0337 ) 0.0236 (0.1518 ) 0.0190 (0.1367 ) 0.0637 (0.2442 ) 0.1390 (0.3460 ) 0.1481

0.0000 (0.0000 ) 0.0019 (0.0436 ) 0.0266 (0.1611 ) 0.0247 (0.1554 ) 0.0564 (0.2308 ) 0.1598 (0.3665 ) 0.1566

0.0000 (0.0000 ) 0.0014 (0.0370 ) 0.0245 (0.1547 ) 0.0208 (0.1427 ) 0.0614 (0.2401 ) 0.1454 (0.3525 ) 0.1507 55

August

September

October

November

December

Panel E: Country of origin Iraq

Afghanistan

Iran

Somalia

Syria

Myanmar

Yugoslavia

BosHz

Serbia

Russia

Country of origin <200 in sample

(0.3183)

(0.3129)

(0.3208)

(0.3051)

(0.3531)

0.0936

0.0913

0.0920

0.0992

0.1572

(0.2912)

(0.2881)

(0.2890)

(0.2991)

(0.3640)

0.1185

0.1072

0.1193

0.1116

0.1836

(0.3232)

(0.3094)

(0.3242)

(0.3150)

(0.3872)

0.1115

0.1014

0.1118

0.1012

0.1875

(0.3148)

(0.3018)

(0.3152)

(0.3017)

(0.3904)

0.0375

0.0317

0.0361

0.0215

0.0612

(0.1900)

(0.1752)

(0.1866)

(0.1452)

(0.2397)

0.0052

0.0079

0.0050

0.0052

0.01

(0.0722)

(0.0887)

(0.0708)

(0.0721)

(0.0997)

0.3113

0.1955

0.3065

0.2278

0.3286

(0.4631)

(0.3966)

(0.4611)

(0.4196)

(0.4698)

0.1901

0.1498

0.1994

0.1567

0.1969

(0.3924)

(0.3569)

(0.3996)

(0.3636)

(0.3977)

0.0683

0.0524

0.0697

0.0542

0.0474

(0.2523)

(0.2229)

(0.2546)

(0.2264)

(0.2125)

0.0532

0.1139

0.0407

0.1717

0.0507

(0.2245)

(0.3178)

(0.1977)

(0.3772)

(0.2194)

0.0355

0.0203

0.0373

0.0124

0.0215

(0.1850)

(0.1409)

(0.1895)

(0.1107)

(0.1450)

0.0458

0.0431

0.0508

0.0209

0.0437

(0.2091)

(0.2031)

(0.2196)

(0.1431)

(0.2044)

0.0406

0.0729

0.0417

0.0457

0.056

(0.1974)

(0.2600)

(0.2000)

(0.2089)

(0.2300)

0.0306

0.0615

0.0304

0.0379

0.0224

(0.1722)

(0.2403)

(0.1716)

(0.1909)

(0.1481)

0.0211

0.0438

0.0225

0.0248

0.0187

(0.1436)

(0.2047)

(0.1482)

(0.1556)

(0.1354)

0.0207

0.0384

0.0199

0.0333

0.0189

(0.1424)

(0.1923)

(0.1395)

(0.1795)

(0.1362)

0.1827

0.2083

0.1812

0.2148

0.1952

(0.3865)

(0.4061)

(0.3852)

(0.4108)

(0.3964)

(0.3552 ) 0.1552 (0.3621 ) 0.1839 (0.3874 ) 0.1898 (0.3922 ) 0.0668 (0.2497 ) 0.0099 (0.0992 )

(0.3636 ) 0.1604 (0.3671 ) 0.1737 (0.3790 ) 0.1833 (0.3870 ) 0.0457 (0.2088 ) 0.0108 (0.1033 )

(0.3578 ) 0.1568 (0.3636 ) 0.1807 (0.3848 ) 0.1878 (0.3906 ) 0.0602 (0.2380 ) 0.0102 (0.1005 )

0.3450 (0.4754 ) 0.2020 (0.4016 ) 0.0492 (0.2162 ) 0.0296 (0.1694 ) 0.0239 (0.1527 ) 0.0492 (0.2162 ) 0.0546 (0.2272 ) 0.0227 (0.1491 ) 0.0190 (0.1367 ) 0.0171 (0.1295 ) 0.1878 (0.3906 )

0.2055 (0.4042 ) 0.1617 (0.3683 ) 0.0380 (0.1914 ) 0.0748 (0.2632 ) 0.0146 (0.1199 ) 0.0438 (0.2046 ) 0.1027 (0.3037 ) 0.0469 (0.2115 ) 0.0425 (0.2018 ) 0.0374 (0.1898 ) 0.2321 (0.4223 )

0.3018 (0.4591 ) 0.1896 (0.3920 ) 0.0457 (0.2089 ) 0.0436 (0.2041 ) 0.0210 (0.1434 ) 0.0475 (0.2127 ) 0.0695 (0.2543 ) 0.0302 (0.1712 ) 0.0263 (0.1600 ) 0.0234 (0.1510 ) 0.2015 (0.4012 ) 56

Number of individuals Number of municipalities

8,400 97

4,292 96

6,947 95

1,532 91

4,282 94

3,519 93

1,577 89

5,096 94

Source: Administrative register information from Statistics Denmark from 19992016. Note: All characteristics refer to the year of assignment. 'Gross sample' refers to 'Gross sample of refugees', 'Balanced panel' refers to 'Balanced panel of household heads', 'Subsample' to ' Subsample of balanced panel of household heads' and 'Estimation samp le' to 'Subsample of household heads and jointly arrived couples'.

Table B3. Summary statistics of municipality of assignment characteristics across samples of refugees. Mean (standard deviation). 3) Subsampl 4) Estimation sample e Date of asylum during the calendar year of asylum: After first 10 municipal After first 10 municipal Any date Any date quotas quotas were filled were filled Men Women Men Women All Men Women All (7) (8) (1) (2) (3) (4) (5) (6) 4.1347 4.0513 4.1623 3.9613 4.1266 4.1696 4.0125 4.121 (1.5910 (1.5658 (1.6084 (1.4740 (1.7056 (1.6356 (1.6856 (1.6784) ) ) ) ) ) ) ) 13.9705 13.706 14.0881 13.4744 14.5838 14.7628 14.0451 14.5407 (7.4552 (7.0556 (7.5743 (6.9702 (8.0071 (7.1303 (7.7528 (7.8798) ) ) ) ) ) ) ) 76.5926 76.6762 76.5643 77.0716 76.8125 76.6978 76.9784 76.7846 (3.9109 (3.8295 (3.9057 (3.6579 (3.7972 (3.7628 (3.7885 (3.7759) ) ) ) ) ) ) ) 46.3182 46.461 46.2382 46.6662 46.0402 45.8603 46.667 46.1099 (8.7216 (8.7530 (8.7645 (8.3664 (9.2213 (9.1865 (9.2172 (9.1746) ) ) ) ) ) ) ) -0.1482 -0.0329 -0.1552 0.0730 0.3422 0.3254 0.3154 0.3223 (1.8772 (1.7481 (1.8797 (1.6223 (1.3893 (1.3696 (1.3831 (1.3785) ) ) ) ) ) ) ) 1.2052 1.2469 1.1699 1.2011 1.127 1.1288 1.0858 1.1155 (1.1334 (1.1589 (1.0359 (1.0748 (0.9021 (0.7992 (0.8717 (0.9206) ) ) ) ) ) ) ) 2.4453 2.541 2.3948 2.5234 2.2585 2.2366 2.2874 2.2524 (1.3132 (1.3023 (1.2295 (1.2307 (1.1415 (1.0602 (1.1171 (1.1523) ) ) ) ) ) ) ) 22.1335 21.7316 22.2939 23.6025 23.5481 23.2501 23.655 23.3754 (25.944 (25.200 (26.221 (24.667 (27.640 (26.254 (27.217 (27.149) ) ) ) ) ) ) ) 17.8545 17.299 17.8957 18.8973 18.7411 18.4587 18.8405 18.5768 (18.205 (17.991 (18.284 (17.988 (18.711 (18.401 (18.615 (18.591) ) ) ) ) ) ) ) 1) Gross sample

Unemployment rate (%)

Unemployment rate of nonWestern immigrants (%) Employment rate (%)

Employment rate of nonWestern immigrants (%) Employment growth (%)

Population share (%)

Non-Western immigrant share (%) Commuting distance to center of local labour market by public transportation Commuting distance to center of local labour market by car

2) Balanced panel

57

Distance to center of local labour market Co-national share (%)

Annual influx of assigned refugees per 1,000 inhabitants

15.0672 14.6565 15.0893 (15.377 (15.341 (15.419 ) ) ) 0.1723 0.1851 0.1639 (0.2178 (0.2443 (0.2103 ) ) ) 0.6575 0.6157 0.6641 (0.4489 (0.4310 (0.4482 ) ) )

Number of individuals Number of municipalities

8,400

4,292 96

97

6,947 95

16.0742 (15.438 ) 0.1861 (0.2238 ) 0.6236 (0.4358 ) 1,532 91

15.8048 (15.665) 0.1440 (0.1925) 0.7339 (0.4983) 4,282 94

15.5222 15.9779 15.6632 (15.704 (15.668 (15.693 ) ) ) 0.1425 0.1401 0.1418 (0.1922 (0.1971 (0.1937 ) ) ) 0.74 0.7088 0.7303 (0.4990 (0.4766 (0.4923 ) ) ) 3,519 93

1,577 89

5,096 94

Source: Administrative register information from Statistics Denmark from 19992016. Note: All characteristics refer to the year of assignment. 'Gross sample' refers to 'Gross sample of refugees', 'Balanced panel' refers to 'Balanced panel of household heads', 'Subsample' to ' Subsample of balanced panel of household heads' and 'Estimation sample' to 'Subsample of household heads and jointly arrived couples'.

Table B4. Summary statistics of employment outcomes for refugees across samples of refugees. Mean (standard deviation). 3) Subsampl 4) Estimation sample e Date of asylum during the calendar year of asylum: After first 10 municipal After first 10 municipal Any date Any date quotas quotas were filled were filled Men Women Men Women All Men Women All (7) (8) (1) (2) (3) (4) (5) (6) 0.3880 0.1613 0.4017 0.1660 0.3341 0.3788 0.1484 0.3075 (0.4851 (0.3555 (0.4614 (0.4873) (0.3678) (0.4902) (0.3721) (0.4717) ) ) ) 1) Gross sample

Employed

By years since asylum: Employed in year 2

Employed in year 3

Employed in year 4

Number of individuals Number of municipalities of assignment

2) Balanced panel

0.3196

0.101

0.3322

0.1044

0.2606

0.3018 (0.4591 ) 0.3992 (0.4898 ) 0.4354 (0.4959 )

0.0818 (0.2741 ) 0.1535 (0.3605 ) 0.2099 (0.4074 )

0.2337 (0.4232 ) 0.3232 (0.4677 ) 0.3656 (0.4816 )

(0.4663)

(0.3014)

(0.4710)

(0.3059)

(0.4302)

0.4021

0.1659

0.4149

0.1697

0.3524

(0.4904)

(0.3720)

(0.4927)

(0.3755)

(0.4778)

0.4439

0.2186

0.4503

0.2239

0.3895

(0.4969)

(0.4134)

(0.4982)

(0.4170)

(0.4877)

8,400

4,292

6,947

1,532

4,282

3,519

1,577

5,096

97

96

95

91

94

93

89

94

Source: Administrative register information from Statistics Denmark from 19992016.

58

Note: All characteristics refer to the year of assignment. 'Gross sample' refers to 'Gross sample of refugees', 'Balanced panel' refers to 'Balanced panel of household heads', 'Subsample' to ' Subsample of balanced panel of household heads' and 'Estimation sample' to 'Subsample of household heads and jointly arrived couples'.

Table B5. Mean (standard deviation) and correlation between municipality characteristics (in percentages) in the 1999-2014-period Correlations Unemployme nt rate

(1) Unemployme nt rate

Unemployme nt rate of non-Western immigrants Employment rate

Employment rate of nonWestern immigrants Employment growth

Population share

Unemployment rate of nonWestern immigrant s (2)

Employm ent rate

Employme nt rate of nonWestern immigrant s

Employm ent growth

Populati on share

Immigra nt share

NonWestern immigra nt share

(3)

(4)

(5)

(6)

(7)

(8)

1.00

Mea n (std. dev.)

(9) 4.19

0.59

1.00

-0.74

-0.28

1.00

-0.44

-0.61

0.31

(1.69 ) 12.1 3 (6.92 ) 75.3 3 (4.31 ) 49.7

1.00

(8.69 ) -0.25

-0.06

0.31

-0.02

1.00

0.05

0.09

-0.16

-0.17

0.15

-0.44 (1.88 ) 1.00

1.02 (1.12 )

Immigrant share

0.04

-0.08

-0.19

0.24

0.004

0.33

1.00

5.63 (3.07 )

Non-Western immigrant share

0.06

-0.01

-0.17

0.22

0.04

0.27

0.95

1.00

3.45 (2.54 )

Number of observations

1,568

Source: Administrative register information from Statistics Denmark from 1999-2014. Note: The table includes information for 98 municipalities over a period of 16 years.

59

Table B6.A. Variable definitions and primary data sources: Individual characteristics. Variable

Definition

Primary data source

Refugee

Dummy for having the residence permit type of refugee. Date of residence permit imputed by the Immigration Service. Dummy for firstarrived adult in the household; if the spouses have arrived on the same date, the husband is defined as the household head. Municipality registered in the population registers in the year of receiving residence permit or the following year. Dummy for being employed.

Residence Permit Register (OPHG), Statistics Denmark (DST).

Education level before immigration, constructed based on an education code of the highest degree attained before immigration. Dummy for source country. Dummy for male. Age calculated as the observation year minus the year of birth observed in the population register. Dummy for married at arrival. Dummy for having a child aged 0-2 years.

Survey-based register on immigrants' educational attainment before immigration, DST.

Dummy for having a child aged 3-17 years.

Population register (BEF), DST.

Date of arrival

Household head

Municipality of assignment

Employed Education level

Country of origin Male Age

Married Child aged 0-2 Child aged 3-17

Residence Permit Register (OPHG), DST.

Residence Permit Register (OPHG) and Population Register (BEF), DST.

Population register (BEF), DST.

Register-Based Labour Force Statistics (RAS), DST.

Population register (BEF), DST. Population register (BEF), DST. Population register (BEF), DST.

Population register (BEF), DST. Population register (BEF), DST.

Table B6.B. Variable definitions and primary data sources: Area Characteristics. Variable

Definition

Primary data source 60

Municipality quota

Population share

Non-Western immigrants share

Co-national share

Unemployment rate

Unemployment rate of non-Western immigrants

Employment rate

Annual maximum quota of refugees to be allocated to the municipality. Number of inhabitants in the municipality in percentage of the total national population. Number of nonwestern immigrants living in the municipality in percentage of the number of inhabitants in the municipality. Number of conationals living in the municipality in percentage of the number of inhabitants in the municipality. Number of unemployed individuals in the municipality in percentage of the labour force of the municipality. Number of unemployed nonWestern immigrants in the municipality in percentage of the non-Western immigrant labour force of the municipality. Number of employed individuals in the municipality in percentage of the number of individuals in working age (18 to 65) in the municipality.

Danish Immigration Service (DIS)

Population register, DST. Authors' calculations based on full population data.

Population register, DST. Authors' calculations based on full population data.

Population register, DST. Authors' calculations based on full population data.

Population register and Register-Based Labour Force Statistics (RAS), DST. Authors' calculations based on full population data.

Population register and Register-Based Labour Force Statistics (RAS), DST. Authors' calculations based on full population data.

Population register and Register-Based Labour Force Statistics (RAS), DST. Authors' calculations based on full population data.

61

Employment rate of non-Western immigrants

Employment growth

Distance time by public transport

Distance time by car

Distance in kilometers Communing area dummies

Number of employed non-Western immigrants in the municipality in percentage of the number of nonWestern immigrants in working age in the municipality. Percentage increase in the employed population compared to the previous year.

Population register and Register-Based Labour Force Statistics (RAS), DST. Authors' calculations based on full population data.

Time distance with public transportation from biggest station (either train or bus station) in the municipality to the central station of the communting area. Time distance by car from the biggest station (either train or bus station) in the municipality to the central station of the communting area (the shortest distance in km). Distance in kilometers. Communtig areas are defined by Statistics Denmark. For a commuting (or Travel to Work) area it holds that i) the majority of the local employed population work in the area and that ii) the majority of the jobs in the area are occupied by people living in the area.

Calculated by using Google Maps. The time distance is calculated at Monday the 12th of March 2018 with arrival time 8 AM.

Population register and Register-Based Labour Force Statistics (RAS), DST. Authors' calculations based on full population data.

Calculated by using Google Maps. The time distance is calculated at Monday the 12th of March 2018 with arrival time 8 AM.

Calculated by using Google Maps. Source: Statistics Denmark (2016), "Færre og større pendlingsområder". URL: https://www.dst.dk/da/Statistik/Analyser/visanalyse?cid=28054

62