The effects of outsourcing on unemployment Jan Mhlmann Joint work - PowerPoint PPT Presentation

The effects of outsourcing on unemployment Jan Möhlmann Joint work with Stefan Groot The Hague, 13 March 2013

Motivation  Outsourcing leads to further fragmentation of the production process (output : value added went from 3.4 to 4.4 in past 20 years)  Allows more specialisation and productivity increases in the long run  Can have distributional effects and cause temporary unemployment

Aims  Analysing the hazard rate of becoming involuntary unemployed, and the probability of finding a new job after becoming unemployed  In particular, we look at the effects of outsourcing (domestic and international) on these hazard rates

Outline  Data and stylized facts ◦ outsourcing survey ◦ unemployment data  Methodology ◦ Cox proportional hazard model  Estimation results ◦ hazard rate of becoming unemployed ◦ probability of finding a new job  Conclusions

Outsourcing survey  Survey on outsourcing, by Eurostat in 2007 - Data  ± 1000 firms (with > 100 employees) - Method - Results  Binary measure: - Conclusions “did you outsource between 2001 - 2006, yes/no?”  Distinction made between domestic and international

Outsourcing shares (2001 – 2006) by size classes (2007) and destination 40% 35% 30% both 25% - Data outsourced internationally 20% 15% outsourced domestically - Method 10% 5% - Results 0% Q1 Q2 Q3 Q4 Q5 - Conclusions About 26% of the firms outsourced some of its activities between 2001-2006

Outsourcing shares (2001 – 2006) by destination - Data - Method - Results - Conclusions

Unemployment data  Tax data (SSB-Banen) contain information on all matches between employers and employees (2000- 2008) - Data - Method  These data can‟t distinguish between voluntary and involuntary unemployment - Results - Conclusions  Therefore we use data on unemployment benefits (SSB- WW)  Merged with jobs based on end-date of job and start- date of unemployment benefits

Outline  Data and stylized facts ◦ outsourcing survey ◦ unemployment data  Methodology ◦ Cox proportional hazard model  Estimation results ◦ hazard rate of becoming unemployed ◦ probability of finding a new job  Conclusions

Cox proportional hazard model  This is a duration model, that explains the hazard rate conditional on the duration - Data  For example: h t is the probability of becoming - Method unemployed shortly after „surviving‟ for period t . - Results  The hazard rate is explained by: - Conclusions ◦ (i) a base hazard rate that varies over time ◦ (ii) a linear function of some explanatory variables

Cox proportional hazard model  Every job (fiscal relation between employer and employee) is an observation  For each job the model needs the duration and failure - Data (yes/no) - Method - Results  A job can end in three ways: ◦ (i) involuntary (failure) - Conclusions ◦ (ii) voluntary (no failure) ◦ (iii) still exists in 2008 (no failure)  A job only affects the estimation results up until the duration of the job

Base hazard rate - Data - Method - Results - Conclusions

Outline  Data and stylized facts ◦ outsourcing survey ◦ unemployment data  Methodology ◦ Cox proportional hazard model  Estimation results ◦ hazard rate of becoming unemployed ◦ probability of finding a new job  Conclusions

Becoming unemployed no outsourcing vs outsourcing - Data - Method - Results - Conclusions

Becoming unemployed no outsourcing vs outsourcing no outsourcing vs domestic - Data - Method - Results - Conclusions

Becoming unemployed 3 2.5 - Data 2 - Method 1.5 - Results 1 - Conclusions 0.5 0 female foreign employee foreign employee outsourced outsourcing (high income (low income internationally domestically country) country)

Becoming unemployed Interaction with wage quartiles (boundaries are 28k, 38k, 54k in 2008 wages): 2 1.8 - Data 1.6 - Method 1.4 1.2 - Results 1 0.8 - Conclusions 0.6 0.4 0.2 0 Q1 (no) Q2 (no) Q3 (no) Q4 (no) Q1 Q2 Q3 Q4 Q1 (int) Q2 (int) Q3 (int) Q4 (int) (dom) (dom) (dom) (dom)

Finding a new job no outsourcing vs outsourcing - Data - Method - Results - Conclusions

Finding a new job no outsourcing vs outsourcing no outsourcing vs international - Data - Method - Results - Conclusions

Finding a new job 1.2 1.1 - Data 1 - Method 0.9 - Results 0.8 - Conclusions 0.7 0.6 female foreign employee foreign employee outsourced outsourcing (high income (low income internationally domestically country) country)

Outline  Data and stylized facts ◦ outsourcing survey ◦ unemployment  Methodology ◦ Cox proportional hazard model  Estimation results ◦ hazard rate of becoming unemployed ◦ probability of finding a new job  Conclusions

Conclusions (i)  Hazard rate for involuntary unemployment seems to follow U-shape - Data  Controlling for duration, risk is higher for women and - Method foreign employees - Results  Different effects for domestic outsourcing (52% more - Conclusions risk) and international outsourcing (32% less risk)  Domestic outsourcing increases hazard rate particularly for lower income groups

Conclusions (ii)  Probability of finding a new job is lower for woman and foreign employees - Data - Method  Former employees of firms that outsourced internationally have about 10% less chance of finding a - Results new job - Conclusions

The effects of outsourcing on unemployment Jan Möhlmann Joint work with Stefan Groot The Hague, 13 March 2013