Immigration and Jobs in Europe Francesco D’Amuri Bank of Italy

October 21, 2016 Migration and European Welfare States conference - Lund University

Francesco D’Amuri (Bank of Italy)

Immigration and Jobs in Europe

Lund University

1 / 36

Disclaimer

The views expressed here do not necessarily reflect those of the Bank of Italy

Francesco D’Amuri (Bank of Italy)

Immigration and Jobs in Europe

Lund University

2 / 36

Aim of this presentation Analyze the impact of immigration on domestic workers in Europe Document how not only how much migration impacts on wages: analysis on employment levels, relative occupation complexity, flows into and out of employment Explore the role of supply in changing the occupational structure. Until now all the research focused on the demand side: IT and off-shoring Acemoglu and Autor, HLE, (2010) Exploit the cross-country institutional variation to test whether the impact of migration depends on labor market flexibility Check whether migration has a different impact during a major labor demand shock (namely, the Great Recession) Lessons for the current situation (refugees?)

Francesco D’Amuri (Bank of Italy)

Immigration and Jobs in Europe

Lund University

3 / 36

Main results

Migration does not have a negative impact on natives’ employment rates Domestic workers move towards relatively more complex jobs: the complexity of newly created jobs is higher relative to the destroyed ones The job upgrade implies a 0.7% increase in wages following a doubling in migrants shares Such positive reallocation process slowed but did not come to a halt during the Great Recession, and it was larger in countries with more flexible labor laws

Francesco D’Amuri (Bank of Italy)

Immigration and Jobs in Europe

Lund University

4 / 36

Background Labor demand is downward sloping (Borjas, QJE, 2003), but most studies fail to find a negative impact of migration on natives’ wages and employment Reasonable hypothesis: migrants complement natives, do not substitute them (D’Amuri et al., EER, 2010; Ottaviano and Peri, JEEA, 2011) Aim of this paper: a look at the black box... What do actually natives do when migrants arrive One of the few studies on Western Europe (apart from Angrist and Kugler, EJ, 2003). See Battisti et al., CESifo, 2014 for an analysis on OECD countries.

Francesco D’Amuri (Bank of Italy)

Immigration and Jobs in Europe

Lund University

5 / 36

The intuition

Rather than considering the labour market as made up of homogeneous and identical workers with potentially different skill levels, we consider that production is divided into a series of tasks that can be organised in a continuum spanning from simple-routine and prevalently manual tasks to complex-interactive and prevalently cognitive tasks. Companies have to perform a range of these tasks in order to produce goods or services; hence the increased supply of some of them may increase the demand for others. If immigrants and native workers specialise in different segments of the task-specialisation spectrum, then more immigrants can generate higher demand for natives.

Francesco D’Amuri (Bank of Italy)

Immigration and Jobs in Europe

Lund University

6 / 36

The intuition

For instance, for a construction company the supply of more construction workers performing manual tasks (such as installing dry-walls and raising foundations) generates the need for more construction supervisors, technicians, engineers, clerks, and sale representatives (as the company grows) who typically perform more interactive and complex tasks.

Francesco D’Amuri (Bank of Italy)

Immigration and Jobs in Europe

Lund University

7 / 36

The intuition

Where C identifies Complex tasks, M Manual tasks and wC and wM are associated wages The only assumption is that Foreigners have a comparative advantage in Manual/Simple taks (or, alternatively, they attach less stigma to them). Francesco D’Amuri (Bank of Italy)

Immigration and Jobs in Europe

Lund University

8 / 36

Data

EU Labor Force Survey (Wester Europe, due to data availability) years 1996-2010 O*NET variables O*NET Variables describing tasks (Peri and Sparber, AEJ:AP 2009). Two groups of tasks: Complex (Communication, Complex, Mental skills) Manual (Manual and Routine skills) Each of the 21 ISCED occupations is attached a task-percentile. A score of 20 means 20 per cent of workers are using that skill less intensively

Francesco D’Amuri (Bank of Italy)

Immigration and Jobs in Europe

Lund University

9 / 36

Descriptive statistics: share of foreign workers

The percentage of foreign born nearly doubles between 1996 and 2010 (in the USA the increase is slightly higher than 30%). Francesco D’Amuri (Bank of Italy)

Immigration and Jobs in Europe

Lund University

10 / 36

Descriptive statistics: relative task complexity

Task complexity increases for home but not for foreign workers (hardly the result of a common demand shock) Francesco D’Amuri (Bank of Italy)

Immigration and Jobs in Europe

Lund University

11 / 36

Descriptive statistics: relative task complexity and immigration

Positive association between foreign born presence and relative complexity of home workers for a given country, education and age Francesco D’Amuri (Bank of Italy)

Immigration and Jobs in Europe

Lund University

12 / 36

Empirical strategy

Collapse data in country (c), education by age (j) cells for each year (t) Estimate   CD = γ · ln(fj,c,t ) + dc,t + dj,t + εj,c,t (1) ln MD j,c,t dc,t country by year fixed effects; dj,t skill by year fixed effects

Francesco D’Amuri (Bank of Italy)

Immigration and Jobs in Europe

Lund University

13 / 36

Endogeneity

Areas with good economic prospects might attract migrants (reverse causality); failure to find negative impact of migration Ideally an exogenous supply shifter is required for identification. Available in exceptional episodes: Mariel boalift (Card, ILLR, 1990), Ethnic germans relocation in Germany (Glitz, JoLE, 2012). Migrants forced to leave country of origin and randomly allocated to another labor market Falling short of such an event, we address endogeneity in three different ways

Francesco D’Amuri (Bank of Italy)

Immigration and Jobs in Europe

Lund University

14 / 36

Endogeneity

First, and less important, in all specifications we define fj,c,t as the share of migrants over total population and not on employment Second, we test whether our results hold during the Great Recession years (2007-2010), when migrants’ share continued to grow while labor demand fell Third, we test the robustness of our results using two different instruments based on Altonji and Card (1991)

Francesco D’Amuri (Bank of Italy)

Immigration and Jobs in Europe

Lund University

15 / 36

Endogeneity: Instruments

Underlying assumption is: new immigrants tend to settle where existing foreign communities are, in order to exploit ethnic networks and amenities. Historical settlements predict subsequent flows, but are not related to present economic conditions IV1: based on LFS data. In each year immigrants allocated to country/skill cells are equal to the total stock multiplied by their distribution in the first year of observation (1996) IV2: based on Census and LFS data. Calculate the distribution of migrants by country of origin in 1991; multiply the number of migrants in each country/skill cell by the subsequent, origin-specific, growth rate. Compared to IV1; PRO: 1991 more distant in the past, exploits info on country of origin - CON: not all countries included

Francesco D’Amuri (Bank of Italy)

Immigration and Jobs in Europe

Lund University

16 / 36

First stage Column Estimates ln(fj,c,t)

ln(fj,c,t)*Young

ln(fj,c,t)*Old

ln(fj,c,t)*Low edu

ln(fj,c,t)*High edu

Coeff Std error Ftest Coeff Std error Ftest Coeff Std error Ftest Coeff Std error Ftest Coeff Std error Ftest

Observations Controls Country by education Education by year Country by year Age by education by year

1 IV1 0.78 [0.016]*** 74.12 0.496 [0.029]*** 23.57 0.669 [0.018]*** 96.62 0.773 [0.018]*** 67.65 0.531 [0.017]*** 36.61 2094

Yes Yes

2 IV2 0.561 [0.035]*** 19.14 0.333 [0.033]*** 17.48 0.614 [0.026]*** 31.2 0.704 [0.032]*** 21.15 0.423 [0.028]*** 32.23 840

Yes Yes

3 IV1 0.86 [0.011]*** 243.15 1.088 0.014*** 178.68 0.779 [0.014]*** 180.81 0.903 [0.012]*** 238.16 0.635 [0.019]*** 42.94 2094

4 IV2 1.043 [0.037]*** 37.46 1.131 [0.034]*** 44.02 0.86 [0.029]*** 80.56 1.076 [0.035]*** 37.51 0.679 [0.032]*** 32.52 840

Yes Yes Yes

Yes Yes Yes

Instruments strongly positive correlated with ln(fj, c, t)

Francesco D’Amuri (Bank of Italy)

Immigration and Jobs in Europe

Lund University

17 / 36

Migration and natives’ relative task performance

Column ln(fj,c,t)

ln(fj,c,t)*Young

ln(fj,c,t)*Old

ln(fj,c,t)*Low edu

ln(fj,c,t)*High edu

Observations Controls Year by Country Year by age by education Country and year

Francesco D’Amuri (Bank of Italy)

1 OLS 0.058 [0.018]*** [0.008]*** 0.033 [0.022] [0.012]*** 0.062 [0.018]*** [0.008]*** 0.065 [0.017]*** [0.008]*** -0.002 [0.024] [0.012] 2106

2 IV1 0.06 [0.021]*** [0.007]*** 0.024 [0.056] [0.020] 0.06 [0.022]*** [0.008]*** 0.065 [0.020]*** [0.007]*** -0.022 [0.039] [0.021] 2094

3 OLS2 0.069 [0.022]*** [0.010]*** 0.041 [0.028] [0.018]** 0.074 [0.022]*** [0.010]*** 0.071 [0.022]*** [0.010]*** 0.03 [0.043] [0.021] 840

4 IV2 0.074 [0.036]** [0.016]*** 0.045 [0.096] [0.054] 0.074 [0.035]** [0.015]*** 0.064 [0.037]* [0.017]*** -0.012 [0.065] [0.042] 840

Yes Yes

Yes Yes

Yes Yes

Yes Yes

Immigration and Jobs in Europe

5 IV1 0.104 [0.011]***

6 IV2 0.076 [0.005]***

205

84

Yes

Yes

Lund University

18 / 36

Is such reallocation taking place only because simple jobs for natives are destroyed?

Francesco D’Amuri (Bank of Italy)

Immigration and Jobs in Europe

Lund University

19 / 36

Migration and natives’ employment rates

Column ln(fj,c,t)

ln(fj,c,t)*Young

ln(fj,c,t)*Old

ln(fj,c,t)*Low edu

ln(fj,c,t)*High edu

Controls Year by Country Year by age by education Country and year

Francesco D’Amuri (Bank of Italy)

1 OLS 0.015 [0.078] [0.013] 0.134 [0.080]* [0.026]*** 0.001 [0.078] [0.015] 0.017 [0.080] [0.013] 0.003 [0.073] [0.015] 2106

2 IV1 0.044 [0.099] [0.018]** 0.341 [0.208] [0.071]*** 0.045 [0.098] [0.023]* 0.047 [0.100] [0.018]*** 0.001 [0.110] [0.031] 2094

3 OLS2 0.028 [0.095] [0.015]* 0.153 [0.089]* [0.033]*** 0.007 [0.097] [0.018] 0.031 [0.097] [0.016]* -0.025 [0.105] [0.024] 840

4 IV2 0.096 [0.156] [0.031]*** 0.181 [0.347] [0.102]* 0.095 [0.154] [0.031]*** 0.081 [0.157] [0.032]** -0.039 [0.219] [0.065] 840

Yes Yes

Yes Yes

Yes Yes

Yes Yes

Immigration and Jobs in Europe

5 IV1 0.11 [0.018]***

6 IV2 0.134 [0.017]***

205

84

Yes

Yes

Lund University

20 / 36

Migration: hirings and separations

Column Estimates Hirings rate

1 OLS

2 IV1

3 OLS2

0.242 0.432 0.196 [0.266] [0.272] [0.325] [0.121]** [0.124]*** [0.158] Hirings’ relative complex/non-complex skill intensity 0.085 0.108 0.088 [0.020]*** [0.021]*** [0.025]*** [0.009]*** [0.009]*** [0.011]*** Separations rate 0.028 0.031 0.066 [0.085] [0.097] [0.091] [0.025] [0.031] [0.028]** Separations’ relative complex/non-complex skill intensity 0.064 0.068 0.069 [0.017]*** [0.020]*** [0.020]*** [0.008]*** [0.010]*** [0.009]*** Observations 1986 1974 840 Controls Country by education Yes Yes Yes Education by year Yes Yes Yes Country by year Yes Yes Yes

Francesco D’Amuri (Bank of Italy)

Immigration and Jobs in Europe

4 IV2 0.587 [0.402] [0.147]*** 0.152 [0.040]*** [0.018]*** -0.046 [0.127] [0.038] 0.102 [0.029]*** [0.017]*** 840 Yes Yes Yes

Lund University

21 / 36

How national institutions (Employment Protection Legislation) interact with this positive reallocation?

Francesco D’Amuri (Bank of Italy)

Immigration and Jobs in Europe

Lund University

22 / 36

Migration impact and EPL

Column EPL index Estimates ln(fj,c,t)

*Low EPL

*High EPL

ln(fj,c,t)* Low edu

*Low EPL

*High EPL

ln(fj,c,t)* High edu

*Low EPL

*High EPL

Observations

Francesco D’Amuri (Bank of Italy)

1 OECD OLS 0.066 [0.020]*** [0.005]*** 0.053 [0.021]** [0.009]*** 0.076 [0.019]*** [0.008]*** 0.054 [0.021]** [0.009]*** 0.022 [0.031] [0.014] 0.038 [0.027] [0.010]*** 1947

2 OECD IV1 0.055 [0.021]*** [0.005]*** 0.047 [0.022]** [0.010]*** 0.06 [0.022]*** [0.008]*** 0.047 [0.022]** [0.010]*** 0.033 [0.038] [0.016]** 0.027 [0.072] [0.020] 1935

Immigration and Jobs in Europe

3 EC89 OLS 0.096 [0.022]*** [0.007]*** 0.028 [0.010]*** [0.005]*** 0.109 [0.022]*** [0.009]*** 0.029 [0.010]*** [0.005]*** 0.016 [0.037] [0.017] 0.02 [0.023] [0.010]** 1220

4 EC89 IV1 0.085 [0.024]*** [0.008]*** 0.019 [0.009]** [0.005]*** 0.096 [0.024]*** [0.010]*** 0.019 [0.009]** [0.004]*** 0.021 [0.041] [0.018] 0.053 [0.063] [0.026]** 1220

Lund University

23 / 36

Migration impact and EPL: Country by Country

Francesco D’Amuri (Bank of Italy)

Immigration and Jobs in Europe

Lund University

24 / 36

What happened during the Great Recession

Francesco D’Amuri (Bank of Italy)

Immigration and Jobs in Europe

Lund University

25 / 36

Migration impact and the Great Recession

Column Estimates ln(fj,c,t)*Before GR

ln(fj,c,t)* GR

Francesco D’Amuri (Bank of Italy)

1 OLS 0.067 [0.017]*** [0.008]*** 0.045 [0.013]*** [0.008]***

2 IV1 0.059 [0.019]*** [0.008]*** 0.038 [0.012]*** [0.007]***

3 OLS2 0.072 [0.020]*** [0.010]*** 0.043 [0.016]** [0.012]***

Immigration and Jobs in Europe

4 IV2 0.08 [0.026]*** [0.014]*** 0.05 [0.025]** [0.020]**

Lund University

26 / 36

Impact on wages: what is the wage gain associated with natives moving to more complex jobs?

Francesco D’Amuri (Bank of Italy)

Immigration and Jobs in Europe

Lund University

27 / 36

Relative complexity and wages

Mincerian equation  ln(wagei,t ) = α + β ∗ ln

Column

Observations Controls Country by education Education by year Country by year

1 log(C/M) 0.115 [0.019]*** [0.009]*** 275608

CD MD

 + dedu,c + dedu,t + dc,t + εi,t i,t

log(C/M)* year 2007 0.117 [0.019]*** [0.018]***

2 log(C/M)* year 2008 0.11 [0.019]*** [0.012]*** 275608

Yes Yes Yes

Francesco D’Amuri (Bank of Italy)

log(C/M)* year 2009 0.117 [0.021]*** [0.014]***

Yes Yes Yes

Immigration and Jobs in Europe

Lund University

28 / 36

Impact of migration on wages

A doubling in the migrants share in a cell implies a 0.7% in average wages 0.007 = 100% ∗ 0.06 ∗ 0.11 0.06=Migration/complexity elasticity 0.11=Complexity/wage elasticity

Francesco D’Amuri (Bank of Italy)

Immigration and Jobs in Europe

Lund University

29 / 36

Conclusions We have shown that migration did not have an adverse impact on the European labor markets during 1996-2010: Native workers moved towards more complex occupations for given employment rates In the process, newly created jobs were more complex than the destroyed ones Such a positive reallocation process was still in place during the Great Recession citizens Reallocation is larger in countries with more flexible labour laws Job upgrade entails a 0.7% increase in average wages for a doubling in migrants’ share Francesco D’Amuri (Bank of Italy)

Immigration and Jobs in Europe

Lund University

30 / 36

Thank you!

Francesco D’Amuri (Bank of Italy)

Immigration and Jobs in Europe

Lund University

31 / 36

The framework

Relative Demand Each labor market (country) is divided into cells of workers with differing observable skills (education and age). We capture this production structure by combining different skill cells in a multi-stage nested Constant Elasticity of Substitution (CES) production function. Output is produced using capital and labor. Labor is a CES aggregate of labor services from workers in different education groups and, in turn, each of those groups is a CES composite of labor services of workers with different ages.

Francesco D’Amuri (Bank of Italy)

Immigration and Jobs in Europe

Lund University

32 / 36

The framework

Relative Demand In each skill (education/age) cell we separate the labor services supplied as complex tasks (C ) and those supplied as simple tasks (S) and consider those inputs as imperfect substitutes, also combined in a CES. We derive the relative demand for complex and simple services for each skill group by equating the ratio of their marginal productivity to the ratio of their compensations.

Francesco D’Amuri (Bank of Italy)

Immigration and Jobs in Europe

Lund University

33 / 36

The framework

Relative Supply Native and immigrant workers divide their labor endowment between simple and complex tasks in order to maximize their utility. Utility depends positively on labor wage and negatively on a stigma associated with simple working tasks. The relative supply of complex tasks increases with its relative compensation and it increases with the relative ability in complex tasks of the group, as well as with its dislike for manual-routine services. We assume that foreigners are less able in relatively complex tasks or that they dislike manual tasks less.

Francesco D’Amuri (Bank of Italy)

Immigration and Jobs in Europe

Lund University

34 / 36

The framework

Equilibrium Equating relative demand and supply we derive the equilibrium relative provision of tasks The relative supply of complex tasks increases with its relative compensation and it increases with the relative ability in complex tasks of the group, as well as with its dislike for manual-routine services. The model predicts a positive impact of the share of foreign-born, f , on the relative supply of complex tasks of natives. Most of the empirical analysis will be devoted to the estimation of such relationship.

Francesco D’Amuri (Bank of Italy)

Immigration and Jobs in Europe

Lund University

35 / 36

The framework

Where C identifies Complex tasks, M Manual tasks and wC and wM are associated wages The only assumption is that Foreigners have a comparative advantage in Manual/Simple taks (or, alternatively, they attach less stigma to them). Francesco D’Amuri (Bank of Italy)

Immigration and Jobs in Europe

Lund University

36 / 36