Career instability in a context of technological change
2 likes121 views
This document discusses the relationship between career instability and technological change, presenting empirical analyses that reveal routine occupational workers face more instability and longer unemployment spells. It utilizes individual-level data from Germany and Great Britain to test hypotheses related to these patterns. The findings indicate a weak link between career patterns and routine-biased technological change, with country-specific variations in unemployment duration and career stability.
Career instability in a context of technological change
- 1. Career instability in a context of technological change Career instability in a context of technological change Lucas Augusto van der Velde Warsaw School of Economics GRAPE Application of Empirical Methods in Modern Economics Poznan University of Economics and Business 2019 Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change
- 2. Career instability in a context of technological change Introduction Motivation Context Over 5 million jobs expected to be automated worldwide (World Economic Forum 2017) New topic in economics Most evidence is on aggregate data (net employment changes) Exceptions: Cortes (2016), Bachmann, Cim and Green (2018). Provide a new empirical test on models’ implications Our contribution Provide first empirical analysis relating career patterns and technological change using individual level data. Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change
- 3. Career instability in a context of technological change Introduction Preview H1 Workers in routine occupations experienced more career instability. Confirmed in UK, less so in Germany Relation appears non-linear H2 Workers leaving routine occupations experienced longer unemployment spells. Confirmed in Germany, less so in UK Unemployed or inactive? Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change
- 4. Career instability in a context of technological change Theoretical considerations Context: Employment changes since 1970’s (US) −.05 0 .05 .1 .15 .2 0 20 40 60 80 100 Skill Percentile (Ranked by Occupational Mean Wage) 1979−1989 100xChangeinEmploymentShare Smoothed changes in employment by occupational skill percentile 1979−2007 Notes: Figure taken from Acemoglu and Autor (2011, pp. 1071) .05 .1 .15 .2 angeinEmploymentShare Smoothed changes in employment by occupational skill percentile 1979−2007 Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change
- 5. Career instability in a context of technological change Theoretical considerations Routine biased technological change Premise: Analyze tasks → units of activity that produce output Task classification: Manual Cognitive / interpersonal Non-Routine Cleaning, repairing Managing, creating Routine Assembling, packing Bookkeeping, spell checking Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change
- 6. Career instability in a context of technological change Theoretical considerations Routine biased technological change Effects of technological progress (Autor et al. 2003, 2006, Acemoglu and Autor 2011) Routine tasks → Substitution effects dominate. → ↓ demand, ↓ price. Non-routine cognitive tasks → Complementarity → ↑ demand, ↑ price. Non-routine Manual tasks → neither complements nor substitutes → ↑ demand, ↑↓ price. Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change
- 7. Career instability in a context of technological change Theoretical considerations How do workers switch tasks Main models → Not considered (e.g. Autor et al. 2003, 2006, Acemoglu and Autor 2011, Goos et al. 2014, Jung and Mercenier 2014) Jaimovich and Siu (2012) → switching market with lower efficiency → Lower efficiency reflects learning skills → Non-routine is an absorbing state Carrillo-Tudela and Visschers (2013) → Switch leads to occ. specific human capital loss → ↑ Pr(unemp) → ↑ Pr(switch) Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change
- 8. Career instability in a context of technological change Method & data Data 1. German Socioeconomic Panel (GSOEP) 1984 - today (West Germany) > 1500 individuals with balanced data (1991-2000) 2. British Household Panel Survey (BHPS) 1991 - 2008 → Discontinued > 2500 individual with balanced data (1991-2000) Descriptives 3. Occupation Network (O*NET) Grouped data from US Applied to EU before (e.g. Goos et al. 2014, Keister and Lewandowski 2016) Can it be justified? (Hardy et al 2018) Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change
- 9. Career instability in a context of technological change Method & data Measuring the task content of jobs Definition follows Acemoglu and Autor (2011). 16 variables from 3 domains (O*NET) Define 5 task types (all standardized) a) Non-routine: Cognitive, Interpersonal, Manual b) Routine: Cognitive, Manual RTI = Non-routine − Routine (also standardized) Discretization of the RTI variable → Quintiles Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change
- 10. Career instability in a context of technological change Method & data Hypotheses H1 Workers in routine occupations experienced more career instability. H2 Workers leaving routine occupations experienced longer unemployment spells. Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change
- 11. Career instability in a context of technological change Hypothesis 1: Career instability Hypothesis 1: Measuring career instability What is instability? Instability: distance from a perfectly stable career. What is a perfectly stable career? Continuous employment in “similar” occupations How does a career look like? Are careers consistent with RBTC? How to measure distance? Optimal matching Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change
- 12. Career instability in a context of technological change Hypothesis 1: Career instability Optimal matching Imagine two workers with careers: W1 E - U - E - E W2 U - E - E - E How to make these careers equal? 1. Substitution → W2: E - U - E - E 2. Insert and Delete (INDEL) → W2: E - U - E - E - ¡E Minimum number of steps ⇒ Optimal matching Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change
- 13. Career instability in a context of technological change Hypothesis 1: Career instability Optimal matching: Finding the right costs Distance depends on how costs are operationalized Bug or a feature? (Wu (2000) & Levine (2000) Minimum criteria “Costs should be symmetric, fulfill the inequality triangle and be 0 only for the substtution of an element with itself” Studer and Ritschard (2016) Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change
- 14. Career instability in a context of technological change Hypothesis 1: Career instability Optimal matching: Finding the right costs Setting substitution costs 1. All costs equal to 1 2. Costs =1/ transition probability (Lesnard 2010) (period specific, computationally intensive, may be nonsensical) 3. Use theory derived measures (Hollister 2009) Against: arbitrary, which dimension is relevant? Setting INDEL costs 1. 0.5 ×maximum substitution cost 2. Costs should be prohibitely high (Lesnard 2010) 3. Costs should depend on context (Hollister 2009) → 0.5 might be optimal → Computationally intensive Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change
- 15. Career instability in a context of technological change Hypothesis 1: Career instability Optimal matching: our approach Definitions Proposals Career elements RTI quintiles+ NE Substitution costs One Differences in RTI + one to/from NE Indel costs Half of max. substitution costs Reference sequence Continuous employment in same element Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change
- 16. Career instability in a context of technological change Results Measures of instability Group 1 2 3 4 5 6 (NE) Germany OM1 - unit cost 0.20 0.32 0.39 0.35 0.33 0.64 *** OM2 - RTI costs 0.13 0.16 0.20 0.18 0.22 0.64 *** Great Britain OM1 - unit cost 0.28 0.36 0.46 0.42 0.39 0.54 *** OM2 - RTI costs 0.22 0.19 0.36 0.25 0.29 0.54 *** Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change
- 17. Career instability in a context of technological change Results Uncovering effect of RTI Specification yt,t+1 = β0 + β1RTIt + controls + t where yt,t+1 is a measure of instability. β1 is coefficient of interest → Hypothesis: β1 > 0. Other controls: year of birth, gender, educational attainment, city. Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change
- 18. Career instability in a context of technological change Results Results Specification: yt,t+1 = β0 + β1RTIt + controls + t Germany Great Britain OM-Unit OM-RTI OM-Unit OM-RTI RTI 0.01 0.01 0.03*** 0.02** (0.02) -0.01 (0.01) (0.01) R2 0.04 0.06 0.03 0.03 N 1593 1593 1985 1985 Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change
- 19. Career instability in a context of technological change Robustness checks Non-linearities Quintile Germany Great Britain OM-Unit OM-RTI OM-Unit OM-RTI 1 Baselevel 2 0.10* 0.01 0.07** -0.03 (0.05) (0.03) (0.03) (0.02) 3 0.13*** 0.03 0.16*** 0.13*** (0.04) (0.03) (0.04) (0.02) 4 0.10** 0.02 0.12*** 0.02 (0.05) (0.03) (0.03) (0.02) 5 0.07 0.05 0.09** 0.06** (0.05) (0.03) (0.04) (0.03) R2 0.05 0.06 0.05 0.08 N 1593 1593 1985 1985 Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change
- 20. Career instability in a context of technological change Robustness checks Robustness checks 1. Alternative measures of instability % not in mode, # elements, # jobs 2. Alternative career elements Occupation grouped based on distance of tasks, employment status Results not affected by these choices Follow our expectations Resilient to robustness checks Statistically significant..., but economically relevant? Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change
- 21. Career instability in a context of technological change Robustness checks Hypotheses H1 Workers in routine occupations experienced more career instability. H2 Workers leaving routine occupations experienced longer unemployment spells. Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change
- 22. Career instability in a context of technological change Hypothesis 2: Unemployment spells Method Specification timeNE,t = f (RTIt−1, controls) where timeNE,i → lenght of non-employment spell i starting in t. f (·) → log-logistic hazard rate. RTIt−1 → RTI last occupation → H0: βRTI > 0. other controls: year of birth, educational level, gender and spell number. Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change
- 23. Career instability in a context of technological change Hypothesis 2: Unemployment spells Results Specification: timeNE,t = f (RTIt−1, controls) Germany Great Britain NE U I NE U I RTIt−1 0.09*** 0.10*** 0.06 0.05 -0.06 0.18*** (0.03) (0.03) (0.04) (0.04) (0.04) (0.05) N 3215 2636 597 2286 966 1284 Mean dur. 22.94 13.40 48.60 39.09 12.25 52.54 Median dur. 8 4 27 11 5 28 Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change
- 24. Career instability in a context of technological change Hypothesis 2: Unemployment spells Results: predicted survival curves Specification: timeNE,t = f (RTIt−1, controls) 0 .2 .4 .6 .8 1 Survival 0 10 20 30 40 50 Months in non−employment RTI Quintile: 1 2 5 Predicted survival curves Germany .2 .4 .6 .8 1 Survival 0 10 20 30 40 50 Months in non−employment RTI Quintile: 1 2 3 4 5 Predicted survival curves Great Britain Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change
- 25. Career instability in a context of technological change Hypothesis 2: Unemployment spells Robustness checks 1. Non-linear relations 2. Alternative survival functions 3. Non-parametrict estimation Results are consistent Follow our expectations Statistically significant..., but economically relevant? Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change
- 26. Career instability in a context of technological change Conclusions Conclusions Weak link between career patterns and RTI Link is country specific 1. Longer unemployment spells in Germany. 2. More unstable careers in Great Britain. How to reconcile empirical results and theory 1. Embedded technological progress. 2. Link human capital loss to differences in task content. Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change
- 27. Career instability in a context of technological change Conclusions Other potential application of Optimal Matching Life-cycle sequences Retirement patterns, school-to-work transitions, etc. Old idea with new data (SHARE, GGP, etc.) Reactions to events Female labor supply around childbirth Death of a relative... Experiments (?) where actions can take place in different order and order matters Evolution of cooperation Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change
- 28. Career instability in a context of technological change Conclusions Thank you for your attention Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change
- 29. Career instability in a context of technological change Bibliography Bibliography I Acemoglu, D. and Autor, D.: 2011, Skills, tasks and technologies: Implications for employment and earnings, Handbook of Labor Economics 4, 1043–1171. Autor, D., Katz, L. F. and Kearney, M. S.: 2006, The polarization of the US labor market, American Economic Review 96(2), 189–194. Autor, D., Levy, F. and Murnane, R. J.: 2003, The skill content of recent technological change: An empirical exploration, Quarterly Journal of Economics 118(4), 1279–1333. Carrillo-Tudela, C. and Visschers, L.: 2013, Unemployment and endogenous reallocation over the business cycle, Discussion Papers 7124, Institute for Study of Labor (IZA). Goos, M., Manning, A. and Salomons, A.: 2014, Explaining job polarization: Routine-biased technological change and offshoring, American Economic Review 104(8), 2509–2526. Hollister, M.: 2009, Is optimal matching suboptimal?, Sociological Methods & Research 38(2), 235–264. Jaimovich, N. and Siu, H. E.: 2012, The trend is the cycle: Job polarization and jobless recoveries, Working paper 18 334, National Bureau of Economic Research. Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change
- 30. Career instability in a context of technological change Bibliography Bibliography II Jung, J. and Mercenier, J.: 2014, Routinization-biased technical change and globalization: Understanding labor market polarization, Economic Inquiry 52(4), 1446–1465. Keister, R. and Lewandowski, P.: 2016, A routine transition? Causes and consequences of the changing content of jobs in Central and Eastern Europe, (05/2016). Lesnard, L.: 2010, Setting cost in optimal matching to uncover contemporaneous socio-temporal patterns, Sociological Methods & Research 38(3), 389–419. Studer, M. and Ritschard, G.: 2016, What matters in differences between life trajectories: A comparative review of sequence dissimilarity measures, Journal of the Royal Statistical Society: Series A (Statistics in Society) 179(2), 481–511. Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change
- 31. Career instability in a context of technological change Bibliography Sample selection Germany Great Britain In Out In Out Age 38.36 38.44 37.64 36.37 *** Female 0.4 0.56 *** 0.55 0.48 *** Education Prim. 0.72 0.74 *** 0.59 0.64 *** Sec. 0.14 0.12 0.28 0.27 Univ. 0.14 0.14 0.12 0.1 Employed 0.94 0.76 *** 0.85 0.84 Occ. codes (1-3) 0.42 0.41 *** 0.44 0.39 *** (4-6) 0.25 0.24 0.3 0.29 (7-9) 0.33 0.35 0.26 0.32 N 1818 2427 2512 2191 Notes: In/out refers to whether observations were included in the final sample (10 year balanced panel with at least one employment spell). *** denotes significant differences at the 1% level from t-test (age) and χ2 tests (all remaining variables). Back Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change
- 32. Career instability in a context of technological change Bibliography Variables used to derive the task content Non-routine Routine Cognitive Analyzing data/information (A) Importance of repeating the same tasks (C) Thinking creatively (A) Importance of being exact or accurate (C) Interpreting information for others (A) Structured v. Unstructured work (re- verse) (C) Interpersonal Establishing and maintaining personal relationship (A)s Guiding, directing and motivating sub- ordinates (A) Coaching/developing others (A) Manual Operating vehicles, mechanized devices, or equipment (A) Pace determined by speed of equipment (C) Spend time using hands to handle, con- trol or feel objects, tools or controls (C) Controlling machines and processes (A) Manual dexterity (Ab) Spend time making repetitive motions (C) Spatial orientation (Ab) Domains: (A): activities , (C): context , (Ab): abilities Back Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change
- 33. Career instability in a context of technological change Bibliography Routine task intensity −3 −2 −1 0 1 2 Routine task intensity (RTI) Germany Examples Min 348 – Religious associate professional p25 610 – Market oriented skilled agricultural worker p50 344 – Government official p75 419 – Office clerk Max 829 – Machine operator Back Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change
- 34. Career instability in a context of technological change Bibliography From RTI to quintiles Germany −3−1.501.53 RTIindex 0 20 40 60 80 100 Skill percentile RTI quintiles: 1 2 3 4 5 Great Britain 3 Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change
- 35. Career instability in a context of technological change Bibliography Sample careers: Germany 0 100 200 300 Individuals 1081 1105 1129 1153 1177 1201 Months since 01/1900 RTI Quintiles 1 2 3 4 5 NE Group 1: Most Non Routine 0 100 200 300 400 Individuals 1081 1105 1129 1153 1177 1201 Months since 01/1900 RTI Quintiles 1 2 3 4 5 NE Group 5: Most Routine Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change
- 36. Career instability in a context of technological change Bibliography Sample careers: Great Britain 0 100 200 300 400 Individuals 1081 1105 1129 1153 1177 1201 Months since 01/1900 RTI Quintiles 1 2 3 4 5 NE Group 1: Most non-routine 0 100 200 300 400 Individuals 1081 1105 1129 1153 1177 1201 Months since 01/1900 RTI Quintiles 1 2 3 4 5 NE Group 5: Most routine Back Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change
- 37. Career instability in a context of technological change Bibliography Do we capture the features of RBTC? Over a 10-year period Individuals tend to remain in their initial element More so in Germany More so in Non-routine occupations Movements to non-employment Higher in more routine occupations Higher in Great Britain Transitions to non-routine occupations Common (GB) Movements among neigbouring groups (DE) Back Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change
- 38. Career instability in a context of technological change Bibliography 10 year transition matrix: Germany Year 2001 1990 1 2 3 4 5 NE 1 0.69 0.07 0.05 0.06 0.03 0.09 2 0.17 0.57 0.07 0.06 0.03 0.09 3 0.13 0.06 0.47 0.10 0.08 0.15 4 0.07 0.08 0.07 0.53 0.09 0.16 5 0.04 0.03 0.10 0.09 0.54 0.20 NE 0.19 0.08 0.25 0.15 0.14 0.20 Messages Main diagonal presents largest values Movements to NE higher for more routine Hirings occured in routine and non-routine Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change
- 39. Career instability in a context of technological change Bibliography 10 year transition matrix: Great Britain Year 2001 1990 1 2 3 4 5 NE 1 0.60 0.08 0.06 0.07 0.06 0.13 2 0.18 0.50 0.05 0.08 0.05 0.14 3 0.12 0.08 0.44 0.11 0.06 0.20 4 0.16 0.11 0.07 0.39 0.12 0.14 5 0.12 0.07 0.05 0.16 0.43 0.17 NE 0.14 0.13 0.14 0.17 0.14 0.27 Messages Main diagonal still larger... .. but less so than in Germany Inverse U shape for moves to NE Back Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change
- 40. Career instability in a context of technological change Bibliography Common career patterns Germany Group 1 Group 5 Sequence Frequency Sequence Frequency 1 46.44 5 35.01 161 8.81 56 12.47 16 5.76 545 5.28 14 2.71 565 4.08 141 2.37 53 3.12 Great Britain Group 1 Group 5 Sequence Frequency Sequence Frequency 1 28.47 5 19.86 16 5.32 56 6.31 161 4.4 565 4.21 121 3.94 54 3.04 1616 2.08 545 2.34 Back Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change
- 41. Career instability in a context of technological change Bibliography Alternative measures of instability RTI Quintiles NE 1 2 3 4 5 Germany % not-mode 0.20 0.32 0.39 0.35 0.33 0.64 *** # Elements 1.76 2.05 2.22 2.20 2.01 2.51 *** # Jobs 3.06 3.20 3.37 3.64 3.41 3.76 *** Great Britain % not-mode 0.26 0.36 0.42 0.42 0.39 0.55 *** # Elements 2.31 2.42 2.50 2.61 2.51 2.94 *** # Jobs 4.85 4.67 4.87 4.96 4.80 5.05 *** Notes: Table presents ANOVA tests for differences in means between people in different groups Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change
- 42. Career instability in a context of technological change Bibliography Alternative measure of instability – regression % not mode # elements # Jobs Germany RTI -0.00 0.00 0.08 SE (0.01) (0.04) (0.09) R2 0.03 0.02 0.02 Great Britain RTI 0.01 0.05 0.00 SE (0.01) (0.05) (0.09) R2 0.03 0.02 0.02 Notes: Table presents regressions of instability measures on task content of jobs . Robust standard errors in parentheses. Controls include gender, age, residence, and education Back Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change
- 43. Career instability in a context of technological change Bibliography Optimal matching based on Labor market status Country RTI quintiles NE 1 2 3 4 5 Germany 0.63 0.75 0.64 0.54 0.46 1.34 *** Great Britain 0.68 0.78 0.92 0.70 0.68 1.48 *** Notes: Table presents ANOVA tests for differences in means between people in different groups Back Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change
- 44. Career instability in a context of technological change Bibliography Non-employment duration – non-linear RTI effects Germany Great Britain NE U I NE U I RTI Quint. 2 -0.17* -0.07 -0.05 0.31*** 0.17 0.26 (0.10) (0.09) (0.14) (0.12) (0.12) (0.16) 3 0.12 0.14 0.18 0.27** 0.08 0.30* (0.10) (0.10) (0.12) (0.12) (0.13) (0.16) 4 0.08 0.14 0.06 0.10 -0.19* 0.44*** (0.09) (0.09) (0.12) (0.12) (0.11) (0.17) 5 0.18* 0.18* 0.16 0.24** -0.06 0.51*** (0.10) (0.10) (0.13) (0.11) (0.12) (0.16) N 3215 2636 597 2286 966 1284 Notes: Log-logistic survival function. Additional controls include gender, education, age, urban and spell number Back Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change
- 45. Career instability in a context of technological change Bibliography Alternative parametric specification NE U I Germany RTI 0.09*** 0.08** 0.10*** 0.08** 0.06 0.10* (0.03) (0.04) (0.03) (0.03) (0.04) (0.05) LL -5880 -5722 -4629 -4488 -766.8 -745.9 AIC 11776 11487 9274 9020 1550 1530 Great Britain RTI 0.07* 0.07* -0.06* -0.04 0.22*** 0.16*** (0.04) (0.04) (0.04) (0.04) (0.05) (0.05) LL -4231 -4129 -1514 -1489 -2221 -2171 AIC 8477 8298 3043 3018 4457 4379 Notes: Log-logistic survival function. Additional controls include gender, education, age, urban and spell number. Second columns includes controls for frailty at the individual level. Back Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change
- 46. Career instability in a context of technological change Bibliography Alternative non parametric specifications NE U I Germany RTI 0.96* 0.94*** 0.95 SE (0.02) (0.02) (0.05) LL -20947 -17016 -2570 AIC 41914 34053 5153 Great Britain RTI 0.95* 1.02 0.90*** SE (0.02) (0.03) (0.03) LL -14104 -5626 -6700 AIC 2,320 11277 13425 Notes: Non-parametric Cox model. Additional controls include gender, education, age, urban and spell number. Back Lucas van der Velde Warsaw School of Economics GRAPE Career instability in a context of technological change