Economic and political integration of refugees in Norway: Challenges for policy

Transcription

1 Economic and political integration of refugees in Norway: Challenges for policy Bernt Bratsberg, Frisch Centre/University of Oslo UCLS conference on Integration på arbetsmarknaden och invandringens ekonomiska konsekvenser, Stockholm, October 8, 2018

2 Motivation Refugee integration is a top issue for governments across developed democracies Vibrant literature on labor market integration of refugee immigrants Limited knowledge on how to improve political and civic integration Voting rights (Ferwerda et al. 2018) and citizenship (Hainmueller et al. 2015)

3 Refugee integration: The labor market, the welfare state, and political participation Today: Refugees, family immigrants, and labor migrants Refugee employment with years since entry in Europe and the Nordic countries Welfare program participation Job loss and welfare generosity Political integration: Using dispersal policy to identify the influence of social networks

4 Employment rates of refugees are lower than those of other immigrants, 2014 EU LFS Source: Dumont et al. (2016) (EC OECD)

5 But catching up with years since arrival, 2008 EU LFS Source: Dustmann et al. (2016)

6 Assessing labor market integration Methodological challenges even for the simplest descriptive patterns Cross sectional snapshot may be deceiving Country of origin, cohort, and period effects Selective outmigration or early retirement; include those who were children at entry? Strength of Nordic studies: Longitudinal register data, track cohorts/individuals over time

7 Immigrant inflows to Norway,

8 Note: Age 20 62, Dec 31st Immigrant population shares,

9 Main source of income: employment or social insurance? Note: , age 20 62, not in education, in Norway end of yr

10 Refugee labor market integration in the welfare state Expect low refugee employment rates at entry Over time, acquisition of skills (language, education), improved networks, removal of barriers yield rising integration Labor market entry also gives entitlement to social insurance programs Prior evidence shows that immigrants are vulnerable to business cycle fluctuations and structural change Not clear that integration process prevails over the long term

11 The labor market integration process empirical analysis Immigrant employment by years since admission Adult arrivals Observe outcomes Samples age 25 62, in Norway end of calendar yr Natives same age range observations

12 Regression model Want to isolate effect of time in the host country, controlling for age, any arrival cohort heteregoneity, country of origin effects, period effects, and other factors such as education.

13 Regression model Suppose the outcome (employment or social insurance as main source of income) of a person j belonging to immigrant group I observed in calendar year t can be represented by the following equation: y X A YSM C E I I I I I I I I jt jt a ajt m mjt c cj e ej t jt jt a m c e t while the outcome for a native is represented by y X A N N N N N jt jt a ajt t jt jt a t s isolate effect of time in the host country, controlling for age, any arrival cohort heteregoneity, country of origin effects, period effects, and other factors such as education.

14 Regression results: Predicted employment differential between immigrants and natives A. Men B. Women Employment difference vs. natives Years since entry Refugee Fam imm Fam Nor New EU Old EU 95% CI Note: Differentials are based on a regression model that controls for educational attainment, whether schooling is acquired in Norway, whether the highest attainment is from Norway, whether education information is missing, local unemployment, and age at entry all interacted with the five admission categories. The regression further controls for age, county of residence, year of observation, and country of birth, as well as educational attainment and local unemployment interacted with native status. Differentials are evaluated at the weighted average educational attainment in each immigrant sample.

15 Is decline in refugee employment mirrored by «social insurance assimilation»?

16 Regression results: Predicted social insurance differential between immigrants and natives A. Men B. Women Social insurance difference vs. natives Years since entry Refugee Fam imm Fam Nor New EU Old EU 95% CI

17 Main social insurance transfer programs Introduction program: Targeted at refugees and their dependents. Offers stable annual income of 2G (193,766 NOK). Social assistance: Means tested income support, primarily targeted at adults with no other mean of economic support. Unemployment insurance: Support program for active job seekers, eligibility conditional on past contributions. Replacement rate 62.4% of earnings last cal yr. Disability insurance: Temporary; permanent. Replacement rates around 66% of presumed forgone earnings. Eligibility requires that work capacity is reduced by at least 50 % from health problem. Other programs: Family related support programs, such as general child allowances and transitional lone parent support.

18 Source of social insurance transfer Note: , age 20 62, not in education, in Norway end of yr, SI main source of income

19 Key lesson The labor market integration process loses steam and goes into reverse after just a few years After shrinking for 5 years, immigrant native employment differentials stabilize and then widen again Refugees as well as family immigrants This suggests that many refugees do not realize their full employment potential There is scope for improvement!

20 Why do immigrant native differentials widen? Bratsberg, Bernt, Oddbjørn Raaum, and Knut Røed (2018a), Job Loss and Immigrant Labor Market Performance, Economica Bernt Bratsberg, Oddbjørn Raaum, and Knut Røed (2018b), Immigrant Responses to Social Insurance Generosity, IZA dp Bratsberg, Bernt, Oddbjørn Raaum, and Knut Røed (2017), Immigrant Labor Market Integration across Admission Classes, Nordic Economic Policy Review

21 Bratsberg, Bernt, Oddbjørn Raaum, and Knut Røed, Job Loss and Immigrant Labor Market Performance, Economica, forthcoming Bernt Bratsberg, Oddbjørn Raaum, and Knut Røed, Immigrant Responses to Social Insurance Generosity, Frisch Centre, November 2017 Bratsberg, Bernt, Oddbjørn Raaum, and Knut Røed (2017), Immigrant Labor Market Integration across Admission Classes, Nordic Economic Policy Review 2017, Her med oppdaterte data

22 Job loss and immigrant labor market performance Studies immigrant and native workers well established on the labor market Tracks their careers over 10 yrs Identification from mass layoffs; bankruptcies and major downsizings Immigrants face twice the risk of job loss And suffer twice the adverse consequences of job loss compared to natives Lower likelihood of finding new job Lower future earnings

23 Immigrant Responses to Social Insurance Generosity

24 Research questions Labor market success of immigrants from low income countries; what is the role of generous social insurance replacement rates? A. Do program participants postpone transition to jobs when program benefits become more generous? B. Are immigrants more responsive to change in benefits than natives? C. Do higher benefits affect future earnings (from work) and incomes? Different for immigrants? D. Immigrant background or other characteristics with differential benefit effects (e.g., earnings potential)?

25 Identification of effects Challenge: Benefit formula contains rich set of individual characteristics => benefits correlate with unobserved factors that affect outcomes => need of rules generating (exogenous) random variation in benefits Solution: TDI program benefit reform of January 2002, new principles, possible to calculate benefits under old and new rules => randomassignment like variation across eligible applicants From a pension model to an earnings replacement model Only earnings the last three years matter Higher minimum levels and benefits for immigrants with few years of residence in Norway Lower child allowances

26 Contributions and results Extends the literature on heterogeneous benefit effects A. Higher benefits postpone the transition from TDI to employment and reduce future labor earnings B. Immigrant responses are significantly stronger than those of natives C. The stronger response among immigrants is partly due to their larger share with low earnings potential (for whom responses are larger even for natives) D. Total after tax income increases among natives in response to higher benefits, but much less so for immigrants E. Find cross effects on spouses labor supply, particularly from immigrant women to immigrant men

27 Policies that promote integration (Refugees educated in Norway do better, BRR 2017) Subsistence requirements for family reunification (BR 2012; BR in progress) Activation requirements for social assistance (BRR 2018c) Introduction program? Settlement location

28 Introduction program The major program for refugee integration since 2004 Refugees and their families who have been granted a residence permit in Norway have the right to/are obliged to complete an introductory program. All municipalities that settle refugees are obliged to offer the program The purpose is to increase the possibility of newly arrived immigrants participating in working and social life and to increase their financial independence Two (three) years, individual stipend/grant of 2G (NOK 193,766) per year Source: IMDI web page

29 Introduction program Who Age Recent refugees and family With needs for basic qualifications 2017: participants What Fulltime training 2 3 yrs Individualized plans Qualifications not education Mandatory program stipend tied to active participation Test requirements since 2013 Pass requirements since 2017 How Transfers from central admin to municipalities 2016 budget: NOK 11.5 bill Municipality autonomy in organisation Legislated requirement, but few guidelines Adapted from Djuve and Kavli (2017) and IMDi (2017)

30 Program takeup rates Treatment grp: Iraq (28%), Somalia (16%), Afghanistan (8%), Kosovo (7%), Iran (6%) Control group: Pakistan (11%), Vietnam (10%), Sri Lanka (8%), Turkey (7%), Iran (5%).. Iraq (4%), Kosovo (4%).. Somalia (2%).. Afghanistan (1%)

31 Introduction program Evaluate effects of program with difference in differences strategy

32 Employment Higher minimum levels N=

33 Program effects on employment Reform effect on employment A. Men B. Women Higher minimum levels Years since entry Estimate 95% CI Note: Regressions control for age (indicators 25 57), local unemployment rate, county of recidence, and year of observation ( )

34 Social insurance

35 Program effects on social insurance Reform effect on social insurance A. Men B. Women Higher minimum levels Years since entry Estimate 95% CI Note: Regressions control for age (indicators 25 57), local unemployment rate, county of recidence, and year ( )

36 Program effects on log earnings Reform effect on log earnings A. Men B. Women Higher minimum levels Years since entry Estimate 95% CI Note: Regressions control for age (indicators 25 57), educational attainment at admission, local unemployment rate, county of recidence (current and at admission; 38 indicators), year ( ) and country of birth (136

37 Refugee resettlement locations, networks, and political integration Bernt Bratsberg, Jeremy Ferwerda, Henning Finseraas, Andreas Kotsadam Frisch Centre internal seminar, Son, June 1, 2018

38 Research question We ask whether initial resettlement location shapes subsequent political engagement Resettlement location found to influence range of integration relevant outcomes Labor market outcomes (Aslund and Rooth 2007, Beaman 2012, Damm 2014, Godøy 2017), crime (Damm and Dustmann 2014), and health (White et al. 2016) Ours first to study political integration

39 Why should placement matter? Characteristics of local networks available to refugees after arrival might be important Labor market integration fosters political integration Social capital in the network spurs civic integration Political participation is a function of habituation and peer effects (Nickerson 2011, Coppock and Green 2016) Initial experiences particularly important for refugees political integration (Ferwerda et al. 2018) Initial period «integration window» Cues from neighbors and peers particularly important

40 Institutional setting Figure: Refugee admissions to Norway by group Own calculations

41 Institutional setting Country of birth, UNHCR resettlement vs asylum refugees Resettled 50, Asylum 106, Total 156, A. Resettled land Freq. Percent Cum Bosnia , Syria 564 5, Iran 456 4, Kosovo 161 4, Iraq 452 4, Myanm 420 2, Afgh 404 2, Viet 575 2, Congo 279 1, Somal B. Asylum land Freq. Percent Cum Somal , Eritr , Syria , Iraq , Afgh 404 9, Kosovo 161 7, Iran 456 4, Russia 140 4, Ethiop 246 3, Sudan 356 2,

42 Institutional setting We study UNHCR resettlement refugees The number of quota refugees admitted to Norway is decided each year by the Norwegian Storting Directly placed in a Norwegian municipality from abroad Voluntary for municipalities to participate in the refugee placement program Allowed to vote in local elections after three years of residency

43 Research design Want to estimate the effects of initial local networks on refugee immigrants' propensity to vote Our data allow us to measure the characteristics of individuals present in the location of each refugee at the time of placement We can measure turnout among refugees and network members in the 2013 parliamentary election and 2015 local elections for almost half of the electorate

44 Research design

45 Data: Digitized voter census Full electorate in a subset of municipalities, since 2013 Turnout data (voted: yes/no), personal identifier, and constituency Here, focus on 2015: 1,7 million electorate members, covering major cities Covers approx 45 percent of electorate, but only 27 (of 428) municipalities Link to demographics, income, education registers

46 Validating the design Challenges to identification: 1. Selection into neighborhoods 2. Selective outmigration/attrition 3. Reflection

47 Validating the design Figure: Correlation between educational attainment of 1991 pop and turnout in three types of neighborhoods by refugee resettlement

48 Validating the design

49 Main results Full sample: all refugees admitted /placement in 27 municipalities/in 2015 digitalized voter census Network= all residents of neighborhood year of admission

50 Main results

51 Other neighborhood covariates

52 Fine grained networks Age cohorts: all refugees admitted /placement in 27 municipalities/in 2015 digitalized voter census Network= residents in neighborhood year of admission +/ 5 yrs of age of refugee Youth (siblings) sample: age at arrival < 18 Network= residents in neighborhood year of admission, same birth year

53 Fine grained networks

54 Fine grained networks II

55 Importance of network turnout

56 Additional analyses Temporal factors Length of initial exposure; years since arrival Selective mobility? Outmigration; native mobility Reflection Robustness; estimates similar if we exclude highly educated refugees/refugees on disability/large families add country of origin FE

57 Conclusion Early experiences within the host country can shape downstream political engagement Our results suggest that the initial location holds more explanatory power than individual socioeconomic characteristics for refugee electoral participation Our fine grained network results suggest that early exposure to politically engaged peer networks is an important mechanism Optimizing placement policies may provide a cost effective means for host governments to enhance integration outcomes

58 extra

59 Employment and earnings growth Employment Real earnings Index relative to base year Years since base year LDC immigrant EEA immigrant Native

60 Registered unemployment around the time of bankruptcy or major downsizing LDC immigrant EEA immigrant Native Reg unemployed during year (%) Years since base year Bankrupt Downsize Stable

61 Employment before and after bankruptcy or major downsizing Employed (%) LDC immigrant EEA immigrant Native Years since base year Bankrupt Downsize Stable

62 Relative earnings before and after bankruptcy or major downsizing Real earnings relative to base year LDC immigrant EEA immigrant Native Years since base year Bankrupt Downsize Stable

63 Distribution of difference in benefits due to TDI reform Density A. Immigrant men B. Immigrant women C. Native men D. Native women Reform benefit gain (1000 NOK) Potential entrants pre-reform Potential entrants post-reform The distribution of change in benefits from the 2002 TDI reform, by gender and immigrant status. Potential claimants with long term sick leave

64 Empirical model Y Z b PRE b POST R (1 R ) b PRE Rb POST u it it PRE i POST i t t i t i it Y: outcome; Z controls include education, age, family char, time and county fixed effects, and local labor market conditions; b is benefits; R = 1 if post reform Sample consists of both pre and post reform program participants; include calculated pre and post reform benefits for both groups to control for selection, coefficient of actual benefits identifies the causal effect (Fevang et al., 2016; Mullen and Staubli, 2016).

65 Data TDI program participants, entry (3 yrs each side of reform) Age 27 to 59 at entry Follow participants through 2014

66 Estimated hazard rate elasticities wrt TDI benefits Men Women Log actual TDI benefit Effect on transition to: Immigrants Natives Immigrants Natives (1) (2) (3) (4) Employment 0.647*** (0.143) 0.311*** (0.068) 0.424*** (0.127) (0.052) PDI (0.180) (0.102) (0.160) (0.086) Unemployment (0.398) (0.250) (0.675) (0.383) Non participation (0.127) (0.091) (0.111) 0.171** (0.084) Number of spells Number of support points in heterogeneity distribution

67 Estimated effects on annual labor earnings of a Euro increase in TDI benefit A. Immigrant men B. Native men Effect estimate, earnings C. Immigrant women D. Native women Years since program entry

68 Immigrants or characteristics of immigrants?

69 Introduction program Issues: 3 yr post period; too short for evaluation? Effect on social assistance takeup?

70 Social assistance takeup Higher minimum levels N=

71 Refugee labor market integration: Lessons from Norway The labor market integration process loses steam and goes into reverse after just a few years, with rising welfare dependency rates with years in Norway Introduction program: Large lock in effects Positive short term effects for refugee men Small, positive long term term effects on economic status of refugee women Otherwise, no discernable long term effects on economic selfsufficiency

72 Validating the design Figure: Density plots of the distribution of refugees across neighborhoods (A), the relationship between neighborhood turnout and refugees' pre arrival education (B)

73 Validating the design

74 Validating the design

75 Importance of neighborhoods