Presentation is loading. Please wait.

Presentation is loading. Please wait.


Similar presentations

Presentation on theme: "“ LABOR MOBILITY OF IMMIGRANTS: TRAINING, EXPERIENCE, LANGUAGE AND OPPORTUNITIES ” By Sarit Cohen Bar-Ilan University and Zvi Eckstein Tel-Aviv University,"— Presentation transcript:

1 “ LABOR MOBILITY OF IMMIGRANTS: TRAINING, EXPERIENCE, LANGUAGE AND OPPORTUNITIES ” By Sarit Cohen Bar-Ilan University and Zvi Eckstein Tel-Aviv University, University of Minnesota and CEPR

2 2 Introduction The transition pattern of immigrants to a new labor market is characterized by high wage growth, fast decrease in unemployment as immigrants first find blue-collar jobs, followed by a gradual movement to white-collar occupations.

3 3 Focus on - Acquisition of local human capital in: training, experience and local language. Data: quarterly labor mobility since arrival of high skilled male immigrants who moved from the former Soviet Union to Israel. Main macro facts.

4 4 Actual Proportions in White Collar, Blue Collar and Unemployment

5 5 Participation in White Collar and Blue Collar Training

6 6 Formulate a dynamic choice model for: blue and white-collar occupations training related to these occupations Unemployment Labor market opportunities are random and are affected by characteristics, past choices and language knowledge. Participation in training is affected by: the mean wage return, the job offer probabilities, preferences and lost of potential wages.

7 7 Main Results The estimated model fits well the main patterns of the labor market mobility. Return to training: white-collar 19%; blue-collar 13%, for 78% of population and zero for the rest. High return to local experience and language, but –conditional on local human capital - zero return to imported schooling. Main return to training is by the increase of 100% of white-collar offer probability.

8 8 Main Results (cont.) Individual welfare gain at arrival from training programs is 1-1.5%. Aggregate growth rate of wages from the availability of the government provided vocational training programs is.85 percent. Main reasons: return to experience is high and utility from participating in training is low (liquidity constraint).

9 9 Table 3: Multinomial-logit on Employment by Occupation and Unemployment VariableWhite- Collar Unemployed constant-4.4424 (0.5034) -0.4753 (0.4804) Hebrew 0.9612 (0.0761) 0.1342 (0.0701) English0.6563 (0.0428) 0.0205 (0.0052) age at arrival0.0331 (0.0212) 0.0332 (0.0190) Schooling0.0031 (0.0212) 0.0332 (0.0190) training in WC0.9421 (0.1153) 0.8183 (0.1658) training in BC-0.2101 (0.1594) 0.9586 (0.1815) experience-0.0046 (0.0100) -0.6807 (0.0233) occupation in USSR 1.4837 (0.1417) 0.2156 (0.1137) Num. Of Obs.5536 Log likelihood-3558.40

10 10 Table 4: OLS Wage Regression Dependent VariableLn hourly wage white-collar occupation Ln hourly wage Blue- collar occupation Cons 1.091 (0.407) 2.122 (0.120) Hebrew0.129 (0.061) 0.050 (0.027) English 0.132 (0.036) -0.011 (0.022) Age at arrival 0.013 (0.005) -0.003 (0.002) Years of schooling 0.021 (0.022) 0.008 (0.006) Training in WC 0.116 (0.079) -0.009 (0.062) Ttraining in BC-0.045 (0.129) 0.056 (0.055) Experience in Israel 0.017 (0.009) 0.024 (0.003) Num. of Obs.132442 R2R2 0.2300.153

11 11 A Dynamic Choice Model Choice set: Work in a White-Collar job (WC) Work in a Blue-Collar job (BC) Training related to White-Collar jobs (WT) Training related to Blue-Collar jobs (BT) Unemployment (UE)

12 12 Utility by Choice: Wage Functions:

13 13 Transition Probabilities are limited by job-offer probabilities and training-offer probabilities: Individual state and characteristics: last period choice r, experience in Israel, occupation in the country of origin, knowledge of Hebrew and English and training.

14 14 The Model 1. UE 2. UE BC 3. UE BC WC BT WT 20. UE BC WC BT WT Quarter Since Migration: Choices: ……. Study Hebrew

15 15 Solution Method The value function

16 16 The model is solved using backward recursion with a finite linear approximated value at the 21’th quarter as function of S i21. We use Monte Carlo integration to numerically solve for the Value Functions and the probability of the choices jointly with the accepted wages. By simulations we show that the model can capture the main dynamic aspects of the labor market mobility as depicted by the figure.

17 17 Estimation Method The model is estimated using simulated maximum likelihood (SML) (McFadden(1989)) Given data on choices and wage, the solution of the dynamic programming problem serves as input in the estimation procedure. All the parameters of the model enter to the likelihood through their effect on the choice probabilities and wages. Wages are assumed to be measured with error. M=2.

18 18 Results Order Fit of labor market states Fit of transitions and wages Estimated parameters Interpretation of types Policy Implications on training

19 19 Actual and Predicted Proportions in Unemployment, Blue-Collar and White- Collar*

20 20 Actual and ML Proportions in White Collar Training

21 21 Actual and ML Proportions in Blue Collar Training

22 22 Fit results The estimated model fits well the pattern but a formal  2 test rejects the fit of the model. The 5’th year (20%)reduction in BC and increase in WC is explained by : Cohort and prior events (~10%); BC to WC transitions as unemployment reach minimum (~10%).

23 23 Table 6: Actual and Simulated Accepted Wages by Tenure and Training WC occupationBC occupation ActualModelObservati ons ActualModelObservations By quarters in Israel 1-421.76614.215410.47510.96864 5-815.06215.5634610.96811.687139 9-1218.86417.3762911.86812.65873 13-1620.44918.7382512.49713.71797 17-2021.52120.0372815.23214.77569 By training No training17.93216.8409611.98512.211402 After training19.98117.8463612.66013.66640

24 24 Table 7: Estimated Wage Function Parameters Wage parametersBCWC Cons. type11.8799**1.6276 Deviation of type2 from type 1 *0.19300.1443- Hebrew*0.1100*0.0964 English*0.0418-*0.1386 Age at arrival0.00008-0.0050 Years of schooling0.00900.0126 Accumulated experience *0.0187*0.0205 Trained in WC type1*0.1908 Trained in WC type 20.0004 Trained in BC type10.1275 Trained in BC type 20.00008 Proportion of type 1*0.781

25 25 Wage Function Results Very large return to local human capital accumulation: Experience – 2% per quarter, Training- 13 to 19 % by Type; Hebrew – 15 to 19%. Conditional on local human capital – no return to imported human capital.

26 26 Table 8: Estimated Job Offer Parameters WC Offer Probability J=1 BC Offer Probability J=2 b01j1-worked in WC at t-1 type 1*15.9966*2.4980- b01j2-worked in WC at t-1 deviation from type 1 0.0053-*1.7338 b02j1-worked in BC at t-1 type 1*2.9737-*14.0431 b02j2-worked in BC at t-1 deviation from type 1 1.1589-0.0082 b03j1- didn't worked at t-1 type 1 *1.7604-*0.4116- b03j2- didn't worked at t-1 deviation from type 1 0.6392*1.3162 b11j-work experience in Israel 1-40.2761-*0.2421 b12j-work experience in Israel >5*0.8935-*0.2707- b2j-training in occupation j*0.94240.2196 b3j – Age of arrival*0.0286-*0.0071- b4j - Hebrew*0.0938-*0.1744- b5 - English*0.2095 b6 – WC=1 in soviet union*0.5554 b7 - first period dummy*0.4881-

27 27 Table 9: Training and Job offer Probabilities (weighted by types) To/FromWCBCWT Experience01-45+01-45+01-45+ WCAfter training0110.0840.1030.066000 No training1110.0690.0850.0540.0.370.037 BCAfter training0.0680.0520.029111000 No training0.0280.0210.0121110.037 UEAfter training0.2540.2060.12400.3500.4030 295000 No training0.1180.0930.0520.3050.3550.2550.037

28 28 Offer Probabilities Large positive effect of training on WC offers and on BC offers Very Low WT opportunities P=0.037 Very low offers for WC from BC and higher, but low from UE.

29 29 Interpretation of Types Type 2 have unobserved characteristics that fit well the Israeli labor market – easily receive offers and do not need training. (22%). Type 1 – need the training to adjust but the cost is high (utility ~ liquidity problem).

30 30 Policy analysis by Counterfactual Simulations Structural estimation enables to simulate the effect of alternative policy interventions on the choice distribution, wages, unemployment and the discounted expected utility (PV). Policy Choices: Case 1: No training is available. Case 2: Only training in blue-collar (BT) is available. Case 3: Only training in white-collar (WT) is available. Case 4: Double the probability to participate in WT.

31 31 Table 12: Predicted Policy Effects on Mean Accepted Wages and Unemployment (4’th and 5’th years) Policy ChangeNo Training is AvailableDouble WT Offer Rate ImmigrantAccepted wage (% )((Change)Accepted wage (% )(Change) WCBCUEWCBCUE BC in USSR schooling=12-1.1- WC in USSR schooling=15-0.8-

32 32 Table 13: The Predicted Annual Effect of Training Availability on Mean Accepted Wages: Percent Change Relative to an Economy without Training* *Percent change of simulated mean accepted wages on the sample, comparing the training at the estimated model to a no training economy. AllWhite- Collar Blue- Collar Year 1 0.070.1460.035 Year 2 0.601.1720.239 Year 3 0.961.5590.318 Year 4 1.221.8830.396 Year 5 1.402.0290.492 All Years 0.851.6050.261

33 33 Aggregate Wage Growth (Social Rate of Return) Aggregate wage growth is increasing overtime due to the permanent affect on job offers to WC. The social rate of return is above 1% mainly due to type 1 accepting WC jobs and type 2 BC jobs. Better process of job sorting. Double WT opportunities has a high (above 3%) social rate of return.

34 34 Table 14: Predicted Policy Effect on the Hourly Present Value (PV) ExperimentBC in USSR, schooling=12 WC in USSR, schooling=15 age at arrival 30 age at arrival 45 age at arrival 30 age at arrival 45 Upon Arrival* 3,371.873,117.303,458.923,203.37 No Training(-1.11) 3,334.58 (-1.47) 3,071.45 (-0.95) 3,425.98 (-1.35) 3,160.24 No WT(-1.11) 3,334.85 (-1.47) 3,071.45 (-0.95) 3,425.98 (-1.35) 3,160.24 No BT (0.00) 3,371.87 (0.00) 3,117.30 (0.00) 3,458.92 (0.00) 3,203.37 Double WT offer (0.96) 3,404.10 (1.24) 3,155.98 (0.84) 3,487.97 (1.16) 3,240.43

35 35 Table 15: Partition of the Gain from Training by Sources ExperimentBC in USSR, schooling=12WC in USSR, schooling=15 age at arrival 30 age at arrival 45 age at arrival 30 age at arrival 45 No training(3,334.58)(3,071.45)(3,425.98)(3,160.24) No return in all sources (3,334.57) 0.00 (3,071.43) 0.00 (3,425.97) 0.00 (3,160.23) 0.00 Return in utility only(3,335.17) 1.6 (3,072.23) 1.7 (3,426.49) 1.6 (3,160.94) 1.6 Return in utility and terminal (3,361.53) 72.3 (3,105.20) 73.6 (3,448.90) 69.6 (3,190.00) 69.1 Return in utility, terminal, job offer (3,371.20) 98.2 (3,116.63) 98.6 (3,458.10) 97.5 (3,202.49) 98.0

36 36 Conclusions The model provided a way to estimate the social and the individual rate of return from alternative training programs. Most of the gain from training is due to increasing WC job opportunities over long time. Large fraction of wage growth is due to occupational mobility, experience and language learning. The return to imported imported human capital is zero conditional on the locally accumulated human capital.

37 37 TableA1. Summary Statistics ObservationsPercentMeanSD Schooling41914.582.74 Age at arrival41938.059.15 White-collar USSR28467.78 Blue-collar USSR12730.31 Did not work in USSR 81.91 Married36386.63 English4191.760.94 Hebrew before migration 5011.9 Ulpan Attendance38692.3 Ulpan completion33279.2 Ulpan Length (months) 3874.61.34 Hebrew1 (first survey) 4192.710.82 Hebrew2 (second survey) 3162.980.83

Download ppt "“ LABOR MOBILITY OF IMMIGRANTS: TRAINING, EXPERIENCE, LANGUAGE AND OPPORTUNITIES ” By Sarit Cohen Bar-Ilan University and Zvi Eckstein Tel-Aviv University,"

Similar presentations

Ads by Google