Are Immigrants skills priced differently? : Evidence from job polarization in France Catherine Lafineur 1 Eva Moreno-Galbis 2, Jeremy Tanguy 3 Ahmed Tritah 3 1 Nice Sophia Antipolis, GREDEG 2 Aix-Marseille Université, AMSE, GREQAM 3 Université du Maine, GAINS FR CNRS TEPP, IRA
0 10 20 30 40 Share in the host country population (%) Motivation The foreign-born are a substantial and increasing share of labor force TUR MEX HUN FIN CZE PRT DNK EST ISL NLD FRA GBR DEU USA ESP NOR BEL SWE IRL AUT CAN NZL ISR AUS CHE LUX 2013 2000 Source: OECD Population Statistics
Motivation The issues Migration policy is at the forefront of a controversial debate : "how many and who?" Two central issues are at stake 1 The impact of immigrants on natives and host country economies 2 The integration (assimilation) of immigrants to host country economy and society To a large extent answers to these questions lie on 1 Immigrants (relative) skills and 2 How these skills are valuated in their host countries These dimensions are the central focus of this paper
Motivation : new roles for skills In the migration literature skills have mostly been viewed through the lens of standard human capital We observe a shift in the allocation of tasks between capital and labor - and between domestic and foreign labor (Autor et al, 2003 ; Autor and Acemoglu, 2011 ; Grossman and Rossi-Hansberg, 2008) The shift triggered a reallocation of skills across tasks Skills are applied to perform various task within occupations Combination of tasks produce occupational output Skill pricing occurs at the task/occupation level To what extent this reshaping of labor demand/supply has contributed to differences in wage dynamics between immigrants and natives?
Our contribution 1 Identify the contribution of job tasks to the immigrants-natives wage performance 2 We decompose immigrants-natives wage changes into 1 a price effect : changes in the price of skills mandated by technology or globalization 2 a quantity effect : changes in the distribution of skills across tasks - sorting effect 3 We characterize the set of immigrants relative skills (comparative advantage) along dimensions that have not, so far, been systematically scrutinized
Data Employment and wages : French Labor Force Survey : 1994-2012 Task measures at occupational level, O*Net dataset (https ://www.onetonline.org/) A score ranking from (0-100) for 41 distinct tasks = degree/intensity to which a type of task is required to perform the occupation Specific occupational classification based on the US SOC 2010 We assume task content of occupations are similar in France and the US We map the French occupations (PCS) to the US SOC 2010
Task definition We follow Autor, Levy and Murnane (2003) and Autor and Acemoglu (2011) : we distinguish 5 task categories Non-routine manual tasks : skilled manual tasks (e.g. technicians, foremen). Routine cognitive and routine manual tasks : repetitive tasks using an identified procedure (e.g. secretaries, accounting offi cers) ; unskilled manual tasks (e.g. handlers, machine operators). Non-routine analytical and non-routine interactive tasks : requiring responsiveness, creativity, decision making and problem solving (e.g. managers) ; tasks requiring communication skills, physical interaction and adaptability. Each occupation has a specific index value for each task category.
FACT 1 - Employment dynamics along the wage distribution 1994-2012 Yearly employment growth ( %, 1994 2012).15.2.25.3.35.4 0 5 10 15 20 Skill vigintiles ( ranked by occupational median native wage in 1994) natives immigrants natives immigrants
FACT 2 - Wage dynamics by nativity group 1994-2012 Change in real hourly wage by skill level Wage change.02 0.02.04 0 5 10 15 20 Skill vigintiles ( ranked by 1994 natives' occupational median wage) natives Fitted values immigrants Fitted values
Why do wage dynamics diverge despite similar employment dynamics? : the price effect Non routine manual task intensity by skill group Routine task intensity by skill group Non routine manual task intensity of jobs 10 12 14 16 Routine task intensity of jobs 12 13 14 15 16 0 10 20 30 40 50 skill percentile ( ranked by 1994 natives' wage 2 percentiles group) Immigrants Natives 0 10 20 30 40 50 skill percentile ( ranked by 1994 natives' wage 2 percentiles group) Immigrants Natives Non routine analytical/interactive task intensity by skill group Non routine analytical and interactive tasks 60 65 70 75 80 85 0 10 20 30 40 50 skill percentile ( ranked by 1994 natives' wage 2 percentiles group) Immigrants Natives
.4.45.5.55 Task intensity.4.45.5.55 Task intensity.4.45.5.55 Task intensity Introduction Data A conceptual framework Econometric approach Results Robustness check Conclusion Why do wage dynamics diverge despite similar employment dynamics? : dynamics of task specialization Non routine manual Routine Non routine analy tical/interactiv e 1995 2000 2005 2010 2015 Year 1995 2000 2005 2010 2015 Year 1995 2000 2005 2010 2015 Year Nativ es Immigrants
Conceptual framework : general setting Roy,1951 ; Heckman et Honoré, 1986 ; Acemoglu and Autor, 2011 Workers are endowed with a bundle of skills {Skill s } S s=1 the productivity of which {γ jst } varies across occupations j At any time period t, the wage potential of a worker is : ln(wage ijt ) = p jt }{{} occupation specific demand shifter + S s=1 γ jst Skill si Workers sort across occupations based on their comparative advantage The dynamics of wages depend on : dynamics of prices (p jt and γ jst ) : price effect dynamics of skill distribution across occupations (or task) : quantity effect (1)
Conceptual framework :illustration with a special case two skills, task=occupation Workers are heterogeneous with respect to their cognitive skills C i and homogenous with respect to their manual skill : The quantity of effi cient units of labor supplied by a workers i into an occupation j is : s ij () = e β M +γ Mt C i for j = M(anual) e β R +γ Rt C i for j = R(outine) e γ At C i for j = A(abstract) Technology is linear, wages are equal to the value marginal product : W ijt = s ijt p jt for j = M, R, A (3) Workers self-select into jobs offering the highest wage : W (2) ijt = arg max {W imt, W irt, W iat } (4) j=m,r,a
A two sector competitive partial equilibrium model The supply side : selection on comparative advantage along the task complexity ladder
Occupations and inequality At any time period individual (log) wages are equal to : W ijt = ln(p jt ) + β j + γ jt C ij (5) Dynamics of wage inequality across occupations depend on changes in ln(p jt ) and γ jt Dynamics of wage inequality within occupations depend on changes in γ jt Isolating price from returns to skills requires an invariant distribution of skills (C ij ) within occupations
From theory to econometrics : the issues We can not observe potential individual wage changes We assume that within occupations more and less skill workers are located at different (but invariant) deciles of the wage distribution Distance between wage deciles within an occupation reflect differences in (relative) skills Changes in (log) wages within a particular decile q in an occupation j is related to changes in return to skill and changes in task price : ω q jt = ω jt + γ jt }{{} γ j0 a j =between }{{} (ω q j0 ω j0) (6) b j =within
Wage changes within and across occupations : illustration
Importance of composition effects Effect of a decrease in the price of routine task Job polarization No general prediction on observed inequality
Empirical methodology : a three-step approach 1 We obtain residual wages in 1994-96 (period 1) and 2010-12 (period 2) for immigrants and natives ln(w it ) = α t + β t age it educ it resid it + γ ct country it + ω it (7) 2 We regress occupation-decile wage change between period 1 and 2 on the initial value of the occupation-decile : ω q j = a j + b j ω q j0 + εq j (8) We recover as many â j and ˆb j as we have occupations for immigrants and natives ˆb j provides a measure of changes in wage gap between more and less skill workers within occupation â j is an occupation specific wage growth component which results from changes in γ j and p j 3 We regress â j and ˆb j on measures of task content of occupations and immigrants status
Dealing with composition effects We adopt a reweighting approach as elaborated by Dinardo et al. (1996) and Lemieux (2002) Within each occupation we kept constant the age*education composition For immigrants duration of residence is also kept constant Occupations are reweighted along these dimensions as in the base period (1994-1996)
1.5 0.5 Between occupation coefficient.2 0.2.4.6.8 Between occupation coefficient Introduction Data A conceptual framework Econometric approach Results Robustness check Conclusion Between and within occupation wage changes A simple model check Cov(a,b)>0 Nativ es Immigrants 3 2 1 0 1 Within occupation coefficient 3 2 1 0 Within occupation coefficient
Between effect.5 0.5 1 Within effect 3 2 1 0 Introduction Data A conceptual framework Econometric approach Results Robustness check Conclusion Between and within occupation wage changes along the occupational wage distribution Between effect Within effect.2 0.2.4 Median wage 1994.2 0.2.4 Median wage 1994 Natives Immigrants
Wage change 2 1 0 1 2 Wage change 2 1 0 1 2 Counterfactual wage changes for immigrants How immigrants wages would have evolved if their occupational skills were rewarded as natives? How immigrants wages will have evolve if their occupational wage growth were similar to that of natives? Within Between 1.5 0.5 1 Initial wage 1.5 0.5 1 Initial wage Wage change Fitted wage change Counterfactual wage change Fitted counterfactual wage change
Immigrants wage growth premium Between-occupation wage change Within-occupation wage change Scenarios Reweighted 1 Reweighted 2 Reweighted 1 Reweighted 2 (1) (2) (3) (4) Immigrant 0.434*** 0.261*** 0.249*** -0.0326 (0.0208) (0.0160) (0.0442) (0.0436) Population composition constant within group over time YES YES YES YES within and across group over time NO YES NO YES Control for duration of residence YES YES YES YES Observations 164 146 164 146 R-squared 0.715 0.558 0.124 0.003 Note : Robust standard errors in parentheses. Statistical significance : p < 0.01, p < 0.05, p < 0.1
Immigrants wage growth premium : the role of tasks Between-occupation wage change Within-occupation wage change Scenarios Reweighted 1 Reweighted 2 Reweighted 1 Reweighted 2 Reweighted 1 Reweighted 2 Reweighted 1 Reweighted 2 (1) (2) (3) (4) (5) (6) (7) (8) Immigrant 0.439*** 0.262*** 0.438*** 0.240*** 0.262*** -0.00658 0.432*** 0.104 (0.0205) (0.0135) (0.0614) (0.0506) (0.0458) (0.0456) (0.158) (0.192) Non-routine analytical-interactive 0.132*** 0.0780** 0.124** 0.0667 0.394*** 0.287** 0.419*** 0.290* (0.0458) (0.0394) (0.0478) (0.0450) (0.131) (0.136) (0.149) (0.154) Routine manual-cognitive -0.149*** -0.171*** -0.159*** -0.180*** -0.292* -0.403*** -0.396** -0.456*** (0.0529) (0.0506) (0.0588) (0.0591) (0.158) (0.153) (0.178) (0.174) Non-routine manual 0.192*** 0.259*** 0.205*** 0.269*** 0.119 0.294*** 0.230* 0.363*** (0.0482) (0.0462) (0.0509) (0.0531) (0.115) (0.101) (0.119) (0.111) Img*Non-routine analytical-interactive 0.0598 0.0830-0.262-0.0519 (0.182) (0.0786) (0.304) (0.308) Img*Routine manual-cognitive 0.0667 0.0549 0.812*** 0.418 (0.156) (0.0922) (0.276) (0.331) Img*Non-routine manual -0.110-0.0737-0.928*** -0.580** (0.181) (0.0914) (0.317) (0.291) Population composition constant within group over time YES YES YES YES YES YES YES YES within and across group over time NO YES NO YES NO YES NO YES Control for duration of residence YES YES YES YES YES YES YES YES Observations 162 144 162 144 162 144 162 144 R-squared 0.766 0.685 0.767 0.687 0.229 0.104 0.266 0.120 Notes : Robust standard errors in parentheses. Statistical significance : p < 0.01, p < 0.05, p < 0.1 Weighted least square using as weight initial period employment in the occupation for each nativity group
Divergent task and skill pricing for immigrants and natives Differences in immigrants and natives wage dynamics is not related to differences in skill return and/or initial task specific skill endowment Though, some evidence that immigrants skills are less rewarded than natives in non-routine manual tasks This may be due to the fact that immigrants have lower endowment in skills intensively used in performing these tasks
Occupational choices unravelling the dynamics of comparative advantage of immigrants and natives Previous results suggests that on average skills are not differently rewarded for immigrants and natives Therefore changes in occupational choices over-time should inform us about relative skill endowment (comparative advantage) of immigrants and natives Systematic difference in comparative advantage should translate into different task choices
Occupational choices : estimation Workers optimal choices (random utility maximization) could be given a conditional logit formulation : q(j i, time = t) = exp[log(p j t) + S γ j sth st] i s=1 J j=1[exp(log(p jt ) + S γ j sthst] i s=1 Occupational choices are related to relative skill endowments {h i st} and task prices We estimate the probability of choosing an occupation as a function of task content of occupations, interacted with period dummy We use estimated coeffi cients to predict changes in the conditional probability to choose an occupation along the distribution of each task index,
Relative comparative advantage in non-routine manual task Changes in the probability to choose an occupation as a function of its non routine manual task intensity 1994 2012 Change in choice probability.05.1.15.2 1 2 3 4 5 Intensity of non routine manual tasks Natives Immigrants Natives composition constant Immigrants composition constant
Relative comparative advantage in routine task Changes in the probability to choose an occupation as a function of routine task intensity 1994 2012 Change in choice probability 0.05.1.15 1 2 3 4 5 Intensity of routine tasks Natives Immigrants Natives composition constant Immigrants composition constant
Relative comparative advantage in non-routine interactive and analytical task Changes in the probability to choose an occupation as a function of non routine cognitive task intensity 1994 2012 Change in choice probability 0.05.1.15.2 1 2 3 4 5 Intensity of non routine tasks Natives Immigrants Natives composition constant Immigrants composition constant
The impact of minimum wages Focus on workers that are the least likely to be affected by minimum wage changes Increasing minimum wages explain almost half of immigrants relative wage performance over the period immigrants are more concentrated than natives around the minimum wage as a consequence their wage dynamics have been more impacted by rising minimum wages Among the population which is the least likely to be affected by the minimum wage, wage changes are better explained by more favorable returns to skills The increasing returns to skill among more skilled immigrants occurs through their relative upward occupational mobility toward occupations with higher returns to skills
Focus on workers that are the least likely to be affected by minimum wage changes Between-occupation wage change Within-occupation wage change (1) (2) (3) (4) (5) (6) Immigrant 0.130*** 0.134*** 0.123* 0.166*** 0.174*** 0.231 (0.0205) (0.0165) (0.0646) (0.0422) (0.0437) (0.154) Non-routine analytical-interactive 0.130** 0.126** 0.313*** 0.324*** (0.0554) (0.0635) (0.108) (0.121) Routine manual-cognitive -0.294*** -0.315*** -0.370*** -0.466*** (0.0760) (0.0890) (0.136) (0.143) Non-routine manual 0.361*** 0.379*** 0.261** 0.351*** (0.0670) (0.0776) (0.103) (0.0986) Img*Non-routine analytical-interactive 0.0222-0.132 (0.0943) (0.274) Img*Routine manual-cognitive 0.141 0.705** (0.125) (0.323) Img*Non-routine manual -0.139-0.727** (0.115) (0.343) Population composition constant within and across group over time YES YES YES YES YES YES Control for duration of residence YES YES YES YES YES YES Observations 134 132 132 134 132 132 R-squared 0.161 0.448 0.452 0.081 0.185 0.216 Notes : Robust standard errors in parentheses. Statistical significance : p < 0.01, p < 0.05, p < 0.1
Conclusion Globally return to skills for immigrants and natives have evolved similarly Immigrants wages have converged to that of natives over the period. The convergence occurs through upward occupational mobility for the more skilled immigrants (more favorable comparative advantage) due to changes in the minimum wages among the less skilled Further inquiry : role of networks selectivity into and out of employment gender specificities