The Skillsnet project on Medium-term forecasts of occupational skill needs in Europe: Replacement demand and cohort change analysis Paper presented at the Workshop on Medium-term forecast of occupational skill needs in Europe: Interim results. 3RD TECHNICAL WORKSHOP, May 7-8 2007, Maastricht, Netherlands Dr. Ben Kriechel ROA, Maastricht University http://www.fdewb.unimaas.nl/roa
Overview Introduction Cohort-component versus alternatives Core methodology Data availability Participation rates Age structure by countries Dealing with migration Conclusions
Cohort-component vs alternative Cohort-component Not an ideal method as it assumes comparability over different samples Better: Panel over several years of representative sample of workers Sufficient size Detail in ISCO, ISCED, NACE, Age, Gender Lowest common denominator: Cohort component based on LFS cross sections.
Core methodology Changes in level of participation (Changes in) participation by gender and age-group Alternatively: (Changes in) overall participation
Core methodology II Estimates of demographic change - Changes in birthrates - Migration patterns
Core methodology III Measure of outflow: - Using cohort-component method: F & = W age, t W W age 1, t 1 age 1, t 1 - It implies that we need data over more than 5 years.
Core methodology IV Estimation of historical flow coefficient based on cohort-component method. Prediction of future flows based on: Predicted flow coefficient by gender, age and occupation (education) Future demographic size of age cohorts Changes in participation
Data availability Main data source: - Micro data of the European Labour Force Survey (Eurostat) - E3ME (Cambridge Econometrics) - Population forecasts (Eurostat)
Data availability Participation rates: Future participation rates are based on the predictions in the model by Cambridge Econometrics Population forecasts: Europop2004 projection (baseline) by Eurostat
Data availability European Labour Force Survey: - 25 countries - At least annual data - Age-cohorts (5 years) - ISCO, ISCED, NACE - Gender
Data availability Country Short 1997 1998 1999 2000 2001 2002 2003 2004 1 Austria AT - - all all all all all all 2 Belgium BE Q2 Q2 all all all all all all 3 Cyprus CY - - Q2 Q2 Q2 Q2 Q2 Q3 4 Czech CZ Q2 all all all all all all all 5 Denmark DK Q2 Q2 all all all all all all 6 Espana ES Q2 all all all all all all all 7 Estonia EE Q2 all all all all all all 8 Finland FI Q2 all all all all all all all 9 France FR Q1 Q1 Q1 Q1 Q1 Q1 all - 10 Germany DE - - - - - Q2 Q2 Q2 11 Greece GR Q2 all all all all all all all 12 Hungary HU Q2 all all all all all all all
Data availability 13 Ireland IE Q2 Q2 all all all all all all 14 Island IS Q2 Q2 Q2 Q2 Q2 Q2 Q2 Q2 15 Italy IT Q2 all all all all all all all 16 Latvia LV - Q2. Q4 Q2. Q4 Q2. Q4 Q2. Q4 all all all 17 Lithuania LT - Q2. Q4 Q2. Q4 Q2. Q4 Q2. Q4 all all all 18 Luxembourg LU Q2 Q2 Q2 Q2 Q2 Q2 Q1, Q2 Q1, Q3 19 Netherlands NL Q2 Q2 Q2 all all all all all 20 Norway NO Q2 Q3 Q4 all all all all all 21 Poland PL Q2 Q2 all all all all all all 22 Portugal PT Q2 all all all all all all all 23 Slovakia SK - all all all all all all all 24 Sweden SW Q2 Q2 Q2 Q2 all all all all 25 Slovenia SI Q2 Q2 all all all all all all
Data availability: LFS Several countries are missing or incomplete: United Kingdom, Germany Some are short in time-dimension: Austria, Cyprus, Estonia, Latvia, Lithuania, Poland, Slovakia, Slovenia Coding changes in several countries over years
Age structure by country AT: 2003 AT: 2004 BE: 2003 BE: 2004 male fem ale 0 100000 200000 300000 0 100000 200000 300000 0 100000200000300000400000 0 100000200000300000400000 AG E CZ: 2003 CZ: 2004 FI: 2003 FI: 2 004 male fem ale 0 100000200000300000400000 0 100000200000300000400000 0 50000100000150000200000 0 50000100000150000200000 AG E
Age structure by country GR: 2003 GR: 2004 IT: 2003 IT: 2004 ma le fe male 0 100000200000300000400000 0 100000200000300000400000 0 5000001000000 1500000 2000000 0 5000001000000 15000002000000 AG E NL: 2003 NL: 2004 PL: 2003 PL: 2004 ma le fe male 0 200000 400000 600000 0 200000 400000 600000 0 200000 400000 600000 800000 1000000 0 200000 400000 600000 800000 1000000 AG E
Age structure by country SK: 2004 PT: 2003 PT: 2004 SK: 2003 ma le fe ma le 0 100000200000300000400000 0 100000200000300000400000 0 50000100000150000200000 0 50000100000150000200000 AG E EE: 2004 CY: 2003 CY: 2004 EE: 2003 ma le fe ma le 0 500010000150002000025000 0 500010000150002000025000 0 10000200003000040000 0 10000 200003000040000 AG E
Age structure by country IS: 2003 IS: 2004 LT: 2003 LT: 2004 ma le fe ma le 4000 6000 8000 10000 4000 6000 8000 10000 0 50000 100000 150000 0 50000 100000 150000 AG E SE: 2003 SE: 2004 ma le fe ma le 50000100000150000 200000250000 300000 50000100000150000 200000250000 300000 0 20000 40000 6000080000 AG E SI: 2003 0 20000 40000 60000 80000 SI: 2004
Age structure by country IS: 2003 IS: 2004 LT: 2003 LT: 2004 ma le fe ma le 4000 6000 8000 10000 4000 6000 8000 10000 0 50000 100000 150000 0 50000 100000 150000 AG E SE: 2003 SE: 2004 ma le fe ma le 50000100000150000 200000250000 300000 50000100000150000 200000250000 300000 0 20000 40000 6000080000 AG E SI: 2003 0 20000 40000 60000 80000 SI: 2004
Example: Age structure Life science & health associate professionals 0 20000 40000 60000 80000 IT: 2003 0 20000 40000 60000 80000 IT: 2004 male female male female
Example: Age structure Life science & health associate professionals 0 5000 10000 15000 DK: 2003 0 5000 10000 15000 DK: 2004 male female male female
Migration Migration to an from EU countries should if possible be tackled Demographic forecasts should in principle include migration, but we should not solemnly rely on it. Alternative, using outside sources on migration streams, and extrapolating those on replacement demand: Sources can at best distinguish between broad age groups
Conclusion Replacement Demand can in principal be estimated using a common methodology across all countries Data availability dictates the use of methodology For some countries the estimates will be less stable given short time-spans of availability Migration remains a difficult issue to tackle