The Impact of Immigration on Natives Wages: Impact Heterogeneity and Product Market Regulation Susanne Prantl Max Planck Institute for Research on Collective Goods, Bonn Institute for Fiscal Studies, London joint work with Alexandra Spitz-Oener, Humboldt University Berlin University College London April 15th, 2010 0 Motivation Evidence on the impact of immigration on natives wages varies across studies. Selected explanations in recent immigration literature: o Varying degree of substitution between natives and immigrants (Borjas/GroggerHanson 2008, Borjas 2009, Card 2009, Manacorda/Manning/Wadsworth 2009, Ottaviano/Peri 2006, 2008) o Changes in production technologies and workers production tasks (Dustman/Glitz 2008, Lewis 2004, 2005, Peri and Sparber 2010) Our starting point: literature linking product market regulation and labor market outcomes o Krueger/Pischke (1997), Gersbach/Schniewind (1998), Blanchard/Giavazzi (2003) o Bertrand/Kramarz (2002), Kugler/Sauer (2005) o Boeri/Nicoletti/Scarpetta (1999), Griffith/Harrison/Macartney (2007), Angrist/Kugler (2003) 1 1
Research question and empirical approach Research question: How does product market regulation influence the link between immigration and natives wages? Empirical approach: o focus on sudden, large migration wave from East to West Germany after the fall of the Berlin Wall and German reunification Large fraction of immigrants in the medium range of the skill distribution (in contrast to recent US experience, Card 2009) East Germans are closer substitutes in production to native West Germans than traditional immigrants (similarities to Great Black Migration in the US, Boustan 2009) information on vocational training in the GDR before reunification provides instrumental variation (Friedberg 2001) o substantial product market regulation: firm entry regulation in German Trade & Crafts Code regulates some occupations, not all o analysis for the West German labor market 2 Main Findings 1. Heterogeneous impact of immigration on natives wages, depending on firm entry regulation. o Negative response of natives wages to immigration per age- occupation group in non-regulated occupations. In regulated occupations this is not the case. o Our findings are consistent with mixed results for economy-wide wage effects across studies for Germany in the 1990s (Bonin 2005, D Amuri/Ottaviano/Peri 2009, Glitz 2007; Büttner/Rincke 2007, Dustman/Glitz 2008) 2. No indication of differential changes in natives production tasks. 3. Variation in the pattern of natives responses across firm size groups consistent with higher rents in regulated occupations, in particular in large incumbents, and these being shared with employees. 3 2
Outline of the talk Background o Firm Entry Regulation o Labor Supply Shock Data and Descriptive Statistics Empirical Model and Results o Group-level Analysis o Individual-level Analysis Summary and Conclusions 4 Entry regulation German Trade and Crafts Code (Handwerksordnung, HWO): individuals who want to start a legally independent firm in one of the regulated markets need a master craftsman certificate ( Meister ) roots of the law go back to 1897, master craftsman degree gained role as firm entry standard in 1935, confirmed after WWII substantial, effective restriction to firm entry (German Monopolies Commission 1996/97, 2001; German Deregulation Commission 1991) 5 3
Entry regulation Acquiring a master craftsman certificate requires: 1. apprenticeship and graduation (2-3 years) 2. work as journeyman and examination (several years) 3. master craftsman (training and) examination (1-3 years) courses offered by yp private institutions (part- or full-time) exam: occupation-specific parts (theory, practical applications) general purpose parts (bookkeeping & law, marketing & pedagogical training for later training activities) certification by regional committee (5 members, 3 of these with master degree from same occupation), not public substantial restriction: high time requirement, direct costs like course fees Effective restriction? micro-data evidence still limited, firm entry regulation reduces entry into self-employment (Prantl/Spitz-Oener 2009) 6 Illustration of entry regulation Selected occupations with firm entry regulation: o printing and bookbinding o glassblowing o smith and locksmith trades o most textile processing o baker and butcher trades o hairdressing Selected occupations w/o firm entry regulation: o copying and paper production o plate glass production o metal welder o textile refinement trade o ice-cream production and fish processing o beautician trade 7 4
Labor supply shock Migration between East and West Germany, 1985-1999 450 400 350 Immigration from the East Emigration to the East Number of Migrants in '000 300 250 200 150 100 50 0 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 Data: Statistisches Bundesamt 8 Characterization of labor supply shock East Germans distinct from traditional immigrants East Germans quite similar to West Germans Reasons: Germans with full political and economic rights, esp. free access to labor market in West Germany and no specific labor market provisions. No language difficulties. Comparably well educated and easy to integrate due to o vocational training system in the GDR and o reunification contract that acknowledged all training degrees from the GDR. East Germans are closer substitutes to West Germans in production than traditional immigrants. 9 5
Data Qualification and Career Survey (BIBB, IAB) Main advantages of this data for our purposes: o detailed information on occupation in which people work and on occupation in which they were trained initially o information on current main residency and on country/region where people spent most of their youth o detailed information on individual characteristics incl. work tasks Independent, representative cross-sections with ~30000 employed individuals per wave: 1985/86: West Germany 1991/92 and 1998/99: East and West Germany 10 Data Analysis for the region of the former FRG: West Germany. Analysis for medium-educated individuals (vocational training or vocational school degree) in accredited occupations with official educational degree ( anerkannte Ausbildungsberufe, BMBF and BIBB). Analysis on level of age-occupation groups: o 21 grouped 3-digit resp. 2-digit occupation classes with firm entry regulation; 22 occupations without that regulation o age classes: 25-29, 30-34, 35-39, 40-44, 45-49, 50-54 and analysis using individual-level data 11 6
Sample construction Main additional selection criteria: o employees from the survey waves 1985/86, 1991/92 and 1998/99 who are 25-54 years old, work 10-75 hours/week, report relevant information o no employees in public sector, non-profit organizations or the mining and quarrying sector o only employees with German nationality and youth spent in Germany Empirical results robust to several tested sample variants. 12 Descriptive Statistics Share of East Germans in the West German Labor Market Medium Education Low Education 1991 1998 1991 1998 Occupations without Firm Entry Regulation Occupations with Firm Entry Regulation 2.83 4.48 2.24 4.23 3.55 4.20 2.84 5.26 Note: In this table we report the share of East German migrants working in the groups of occupations with and without firm entry regulation, respectively, in the West German labor market. 13 7
Empirical Model: Starting Point w w ( p p ) O A jgt jg ( t ) jgt jg ( t ) jt gt jgt with j : occupation index, g: age index, t: time index w : log real hourly wages of native West German employees averaged within age-occupation-time groups O : occupation-time interactions A : age-time interactions η : error term p jgt E N jgt jgt where E : number of medium-educated East Germans in group jgt in West Germany N : total number of medium- educated employees in group jgt in West Germany and alternatively p jgt E jgt 14 Empirical Model: Main Group-Level Specification w w P NR P R O A NR R jgt jg ( t ) jgt j jgt j jt gt jgt with P p p jgt jgt jg ( t ) R j : dummy variable indicating occupations with firm entry regulation NR j : dummy variable indicating occupations without firm entry regulation and all other variables and parameters defined as before 15 8
Endogeneity Sources for potential bias: 1. Native workers in receiving region may respond to the inflow of labor from outside by migrating themselves. 2. Migrants may self-select into certain occupations and employer decisions i may be endogenous in wage equations. Addressed as follows: Ad 1.: age-occupation group analysis switching occupations is costly due to educational decisions taken early in life Ad 2.: first differencing; i occupation-time and age-time effects; IV approach to address potential self-selection of East German migrants based on time-varying age-occupation-specific wage growth in West Germany 16 Instrumentation (1) Use of instrumental variation following from fall of Berlin Wall and German reunification: I jgt A jgt where A: number of medium-educated East Germans in West Germany with vocational training in occupation j change in I interacted with indicator for regulated resp. nonregulated occupations I likely to be partially correlated with p due to occupation-specific human capital I likely to be is uncorrelated with error term η for 2 reasons 17 9
Instrumentation (2) Reasons for uncorrelatedness between I and η: 1. Fall of Berlin Wall and breakdown of GDR came unexpectedly, migration effectively impossible beforehand East Germans didn t take into account earnings potential in FRG when deciding on vocational training in GDR 2. Planned economy system in the GDR restricted individuals educational choice: o constitution: everybody has the right but also the duty to get a vocational degree (through vocational training, university or technical college) o o political aim of vocational education system: increase social equality state-secretary for vocational training determined the number of training positions per occupation (aim: fulfilling the 5-year production plans) 18 Empirical Model: Main Individual-Level Specification w p NR p R O A S NR R ijgt jgt j jgt j jt gt jg ijgt with i : index for individuals S : age-occupation interactions R j : dummy variable indicating non-regulated occupations NR j :dummy variable indicating regulated occupations and all other variables and parameters defined as before 19 10
Descriptive Evidence All Occupations -.5 cha nge in log mean real hourly wa 0 ages.5 -.3 -.2 -.1 0.1.2 change in share of East German immigrants 20 Descriptive Evidence Occupations w/o Firm Entry Regulation Occupations with Firm Entry Regulation -.5 -.5 change in log mean real hourly wage 0 es.5 change in log mean real hourly wage 0 es.5 -.3 -.2 -.1 0.1.2 change in share of East German immigrants -.3 -.2 -.1 0.1.2 change in share of East German immigrants 21 11
Descriptive Evidence.4 Occupations w/o Firm Entry Regulation.4 Occupations with Firm Entry Regulation change in log mean real hourly wages -.2 0.2 s change in log mean real hourly wages -.2 0.2 -.4 s -.1 -.05 0.05.1 change in share of East German immigrants -.4 -.1 -.05 0.05.1 change in share of East German immigrants 22 Results: Group-Level Analysis The Impact of Immigration from East Germany on Native West Germans Wages Age-Occupation-Level Analysis, OLS and Second Stage Results Dependent Variable: Change in mean log real hourly wages PANEL A OLS IV OLS IV (1) (2) (3) (4) Change in fraction of East Germansaot -0.199 (0.151) -0.341 (0.410) Non-regulated occupations* Change in fraction of East Germansaot -0.469** (0.203) -1.512*** (0.549) Regulated occupations * 0124 0.124 0.444 Change in fraction of East Germansaot (0.198) (0.386) Age group-time effects Yes Yes Yes Yes Occupation-time effects Yes Yes Yes Yes R 2 0.396 0.393 0.404 0.339 F-test: ß NR =ß R (p-value) 4.40 (0.04) 9.31 (0.00) PANEL B OLS IV OLS IV (5) (6) (7) (8) Change in number of East Germansaot -0.009*** (0.002) -0.004 (0.004) Non-regulated occupations* Change in number of East Germansaot -0.012*** (0.003) -0.010*** (0.004) Regulated occupations * 0.002 0.011 Change in number of East Germansaot (0.005) (0.009) Age group-time effects Yes Yes Yes Yes Occupation-time effects Yes Yes Yes Yes R 2 0.408 0.403 0.414 0.410 F-test: ß NR =ß R (p-value) 5.43 (0.02) 5.22 (0.02) Number of observations 495 495 495 495 Note: Robust standard errors in parentheses are robust and allow for correlation within age-occupation groups. 23 12
Results: Group-Level Analysis The Impact of Immigration from East Germany on Native West Germans Wages Age-Occupation-Level Analysis, First Stage Results Dependent Variable: PANEL A Chg. in fraction of EGaot Chg. in fraction of EGaot*NR Chg. in fraction of EGaot*R (2) (4) (4) Change in East Germans (EG) with relevant trainingaot 0.009*** (0.002) Non-regulated occupations (NR)* Change in EG with relevant trainingaot 0.006*** (0.002) 0.001 (0.001) Regulated occupations (R)* Change in EG with relevant trainingaot -0.001 (0.001) 0.013*** (0.002) Age group-time effects Yes Yes Yes Occupation-time effects Yes Yes Yes R 2 0.321 0.311 0.446 F-test, excluded instruments (p-value) 33.52 (0.00) 8.53 (0.00) 25.29 (0.00) PANEL B Chg. in number of EGaot Chg. in number of EGaot*NR Chg. in number of EGaot*R (6) (8) (8) Change in EG with relevant trainingaot 0.757*** (0.116) NR*Change in EG with relevant trainingaot 0.899*** (0.161) 0.001 (0.019) R*Change in EG with relevant trainingaot t -0.049049 0566*** 0.566 (0.044) (0.091) Age group-time effects Yes Yes Yes Occupation-time effects Yes Yes Yes R 2 0.697 0.748 0.637 F-test, excluded instruments (p-value) 42.49 (0.00) 15.87 (0.00) 19.50 (0.00) Number of observations 495 495 495 Note: Robust standard errors in parentheses are robust and allow for correlation within age-occupation groups. 24 Why do we see differential reactions of natives wages to immigration in both groups of occupations? 1. changing composition of workers, differential employment trends, other industry trends? 2. rents and rent-sharing in regulated occupations? 3. differential changes of natives production tasks? 25 13
Results: Individual-Level Analysis Dependent Variable: Log real hourly wages PANEL A OLS IV OLS IV (1) (2) (3) (4) Non-regulated occupations (NR)* Fraction of East Germansaot -0.394*** (0.149) -0.946** (0.482) -0.352** (0.138) -0.834** (0.423) Regulated occupations (R)* Fraction of East Germansaot 0.045 (0.175) 0.405 (0.546) 0.059 (0.177) 0.370 (0.532) NR*Log employment of native WGaot 0.032 0.027 (0.020) (0.021) R*Log employment of native WGaot 0.011 (0.194) 0.011 (0.020) Time-invariant individual characteristics (age, - - Yes Yes age 2, male, tenure, tenure 2 ) Time-varying individual characteristics (age, Yes Yes - - age 2, male, tenure, tenure 2 ) Industry-time effects - - Yes Yes Age group-time effects - - Yes Yes Occupation-time effects Yes Yes Yes Yes Age group-occupation effects Yes Yes Yes Yes R 2 0.323 0.323 0.337 0.336 F-test: ß NR =ß R (p-value) 359(006) 3.59 (0.06) 342(006) 3.42 (0.06) 330(007) 3.30 (0.07) 343(006) 3.43 (0.06) NR, First Stage: F-test: excl. instruments 14.31 12.34 R, First Stage: F-test: excl. instruments 21.05 22.22 Number of observations 17,877 17,877 17,877 17,877 Note: Robust standard errors in parentheses are robust and allow for correlation within age-occupation groups. 26 Results: Wage Response in Small and Large Firms The Impact of Immigration from East Germany on Native West Germans Wages in Small and Large Firms Individual-Level Analysis, OLS and Second Stage Results Dependent Variable: Log real hourly wages PANEL A OLS IV (1) (2) Small Firm*NR*Fraction of East Germansaot -0.611*** (0.191) -0.545 (0.556) Large Firm*NR*Fraction of East Germansaot -0.124 (0.173) -1.637** (0.785) Small Firm*R*Fraction of East Germansaot -0.250 (0.235) -0.455 (0.620) Large Firm*R*Fraction of East Germansaot 0.406* (0.241) 0.935 (0.756) Small Firm*NR -0.085*** (0.011) -0.064*** (0.021) Small Firm*R -0.079*** (0.013) -0.123*** (0.020) R 2 0.349 0.333 F-test: ß L,R =ß S,R (p-value) 4.32 (0.04) 2.74 (0.09) F-test: ß L,R =ß L,NR (p-value) 3.00 (0.08) 5.70 (0.02) F-test: ß L,R =ß S,NR (p-value) 10.90 (0.00) 2.43 (0.11) F-test: ß S,NR =ß L,NR (p-value) 3.93 (0.05) 1.58 (0.21) SNR SR F-test: ß S,NR =ß S,R (p-value) 1.45 (0.23) 0.01 (0.91) F-test: ß L,NR =ß S,R (p-value) 0.18 (0.69) 1.38 (0.24) Small Firm*NR, First Stage: F-test: excl. instr. 12.34 19.64 Large Firm*NR, First Stage: F-test: excl. instr. 12.34 11.14 Small Firm*R, First Stage: F-test: excl. instr. 22.22 8.53 Large Firm*R, First Stage: F-test: excl. instr. 22.22 20.34 Number of observations 17,877 17,877 Notes: Robust standard errors in parentheses are robust and allow for correlation within age-occupation groups. All regressions include time-invariant individual characteristics (as listed above), industry-time effects, age group-time effects, occupation-time effects and age group-occupation effects. 27 14
Results: Production Task Changes? The Impact of Immigration from East Germany on Native West Germans Production Tasks in Small and Large Firms Individual-Level Analysis, OLS Results Dependent Variable: PANEL A Analysic Interactive Manual OLS OLS OLS (1) (2) (3) Small Firm*NR*Fraction of East Germans aot -0.018 (0.121) 0.097 (0.148) 0.317 (0.224) Large Firm*NR*Fraction of East Germans aot -0.143 (0.112) 0.023 (0.144) 0.111 (0.224) Small Firm*R*Fraction of East Germans aot -0.169 (0.186) -0.090 (0.177) 0.031 (0.254) Large Firm*R*Fraction of East Germans aot 0.145 (0.290) 0.083 (0.223) -0.070 (0.353) Small Firm*NR 0.002 (0.006) 0.021*** (0.007) 0.022** (0.011) Small Firm*R -0.010 (0.013) 013) 0.025*** (0.009) 009) 0.016*** (0.015) 015) R 2 0.139 0.498 0.477 Number of observations 17,843 17,843 17,843 Notes: Robust standard errors in parentheses are robust and allow for correlation within age-occupation groups. All regressions include time-invariant individual characteristics (as listed above), industry-time effects, age group-time effects, occupation-time effects and age group-occupation effects. 28 Summary and Conclusions 1. Heterogeneous impact of immigration on natives wages, depending on firm entry regulation. o Negative response of natives wages to immigration in the group of non-regulated occupations. In regulated occupations this is not the case. o Our findings are consistent with mixed results for economy-wide wage effects across studies for Germany in the 1990s. 2. No indication of differential changes in natives production tasks. 3. Variation in the pattern of natives responses across firm size groups consistent with higher rents in regulated occupations, in particular in large incumbents, and these being shared with employees. 29 15