EU 6th framework, EUKLEMS WP 8: Labour Markets and Skill Formation 1) Modelling the Demand for Heterogeneous Labour 2) Background paper: The Impact of International Outsourcing on Employment: Empirical Evidence from EU-Countries Martin Falk and Yvonne Wolfmayr WIFO WP 8: Labour Markets and Skill Formation Stylized facts: • Input factors such as information and communication technologies, imported materials, purchased services, skilled labour as well as general capital have been increasingly used in production • Relative demand for low skilled labour decreases faster over time than does supply • Decline in the demand for older workers • Stable relative wages in some EU countries; rising skill premium in UK, US etc. • Input prices of imported materials grew at a smaller rate than the wages WP 8: Labour Markets and Skill Formation Research questions: • Relationship between output and employment by skill level and age • Labour-labour substitution and wage elasticities of different types of workers • Capital-Skill complementarity • Skill-biased technology change – Impact of R&D capital on heterogeneous labour – Impact of information technology capital on heterogeneous labour • Impact of outsourcing on labour demand and productivity – Outsourcing of services – Outsourcing of production • The labour demand for older workers („age-biased technological change“) WP 8: Labour Markets and Skill Formation • Under cost minimization: Cost function is a function of input prices, fixed factors and output C c(Y ,W11 ,....Wnm , K , T ) • Translog cost function: ln C 0 i ln Pi k ln K yY T T i 1 N N N N 1 ij ln Pi Pj ßik ln Pi ln K ßiy ln Pi ln Y 2 i 1 j 1 i 1 i 1 N ß i 1 it ln Pi ln T WP 8: Labour Markets and Skill Formation • Factor cost share equations ln C * ln C * ln C S ; S2 ; SN ln P1 ln P3 ln PN * 1 • Joint estimation of cost function and factor demand – => identification of the productivity and factor demand effects • Indicators of technological change – R&D capital stock, R&D intensity, R&D spillovers – Infomation technology capital stock – High-Tech capital stock (electrical equipment, instruments...) • In case of two types of labour, the estimation equation becomes: W1 S1 1 ln 2 ln K 3 ln Y 4 ln T W2 WP 8: Labour Markets and Skill Formation Disaggregation of labour • Educational qualification (i) Compulsory school (isced 0-2) (ii) Higher general secondary school (isced 3a), Apprentice training (isced 3b), Higher technical and vocational college (isced 4ab) (iii) academic degree, university degree and post graduates • Age and gender Estimation Problems: • Estimation of elasticities of substitution become impracticable when the number of factors in the system is large • Multicollinearity caused by very high correlation between wage for different skill levels • Use of interpolated data • Dynamic specification, adjustment costs WP 8: Labour Markets and Skill Formation Previous literature • Technology and skills, Cross-country studies – Hollanders, ter Weel (2002): manufacturing four EU countries, white collar high and low-skilled, blue-collar high and low skilled, foreign and domestic R&D capital stock – Machin and Van Reenen (1998), blue/white collar – O‚Mahoney, Robinson and Vecchi (2004): educational qualification, IT and general capital – Morrison-Paul and Siegel (2001) – numerous studies studies for individual countries, see Autor, Katz and Krueger (1998) WP 8: Labour Markets and Skill Formation • Labour-Labour substitution and own-wage elasticities – Hamermesh (1993): own-wage elasticities (in absolute terms) decrease with the skill level – Mellander (2000), four educational qualification groups, Swedish manufacturing data, 20 industries: the own-wage elasticity decreases with the skill level – Riley and Young (1999), five educational qualification groups, U.K. industry panel data mixed results for the ranking of own-wage elasticities by skills • Capital-Skill complementarity: Krusell, Ohanian, Rios-Rull and Violante (2004) – capital-embodied technological change alone can account for most of the variations in the skill premium WP 8: Labour Markets and Skill Formation Previous literature – use of quality-adjusted prices for a number of durable equipment categories such as office and computing equipment including peripheral equipment and accounting machinery (OCAM), communication equipment, general industrial equipment and transportation equipment. – There has been a strong decline in the relative price of equipment (ratio of the price index for capital equipment and the price index for consumption of non-durables and services) of about 7 percent per year – Strong increase in the stock of equipment. • International Outsourcing and demand for skills – Feenstra and Hanson (1999) for the US, Anderton and Brenton (1999), Hijzen, Görg and Hine (2004) for the UK, Geishecker (2002) for Germany; Strauss-Khan (2003), Egger and Egger (2001) for Austria…. • NEXT STEPS: Survey paper Background paper WP 8: The Impact of International Outsourcing on Employment Motivation: • Imported materials are one the fastest growing input factor used in production • Imports from low income Central and East European and East Asian countries most dynamic component of trade • Effects of outsourcing – employment losses – negative distributional effects – productivity gains – gain in competitiveness and market position • Aim of the paper: – New insights into the employment effects of international outsourcing. – Extension of previous work: (i) cross-country study, (ii) Disaggregation:imported materials from high and low wage countries (iii) robustness checks Outline • Previous literature • Empirical model and hypotheses • Data and descriptive statistics • Estimation results • Conclusions Previous literature • Huge literature on the impact of outsourcing on skilled and unskilled workers; in this study: total employment • Negative correlation between employment growth and imports/import prices (Sachs and Shatz, 1994; Greenaway et al., 1999; Revenga, 1992). • Sachs and Shatz (1994): Industry employment levels fall due to imports from developing rather than developed countries. • Neven and Wyplosz (1996): Imports from developing and developed countries have similar effects • Landesmann, Stehrer and Leitner (2001): – import penetration from emerging countries have a significant negative effect on employment growth in the period 1982-1988; effect disappears in the 1990s. – Effect is stronger in the high-skill intensive industries than in the low-skill intensive industries Empirical model and hypotheses • Labour demand model ln Lit ß0 ß1 ln Yit ß2 ln WPit ß3 ln IMQit t i it . – Lit: total employment – Yit: value added in constant prices – WPit: real wage – IMQit: imported materials from the same industry as a percentage of gross output • Estimation equation: ln Li 0 1 ln Yi 2 ln WPi 3IMQi i ∆: average annual change of the variables between 1995-2000 • Estimation methods: (i) OLS using first differences, (ii), robust regression, (iii) median regression (iv) weighted OLS with employment shares as weights Research questions • Impact of imported materials on employment • Impact of imported materials from low-wage and highwage countries on employment • Heterogeneity across industries: (i) two broad industry groups : NACE 29-35 and NACE 15-28; 36 (ii) Declining and expanding industries Data • • Input-Output Table 1995 and 2000 (Eurostat) – imported intermediates – 7 EU countries (Aut, Dk, Fl, G, I, NL, Sw) – NACE 2-digits; manufacturing – No regional breakdown of material imports Definition of outsourcing – narrow measure: purchases from within the same industry class – imported intermediates as percent of gross output • UN Foreign Trade Database: High wage – Low wage countries • OECD STAN Data Descriptive statistics • Share of imported intermediates in gross production: 8.8% (7.2% high wage countries; 1.6% low wage countries). • Strong increase of the share of imports of intermediates from lowwage countries (+9% p.a.) • Kruskal-Wallis test: – high outsourcing industries subject to significantly higher negative total employment growth than low outsourcing industries – Employment losses in these sectors are significantly higher if inputs are sourced from low-wage countries. Data: high wage countries in 2000 7 EU nd Sw s ed en rla et he Ita ly Low-wage countries High-wage countries Au st ria D en m ar k Fi nl an d G er m an y 16 14 12 10 8 6 4 2 0 Imported materials (from the same industry) in low wage and N • Growth of Outsourcing 1995-2000 Average annual percentage change 16 High-wage countries 14 Low-wage countries 12 10 8 6 4 2 0 7 EU ed en he rla et Sw nd s ly Ita N G er m an y nd Fin la ar k en m D A us t ria -2 Summary statistics Mean Q50 Q25 Q75 Std. Dev Min Max all manufacturing industries (# of obs: 144) Average annual growth rate between 1995 and 2000 (%): Value added in constant prices 3.3 2.3 0.0 Total employment -0.8 -0.4 -2.2 Real wages 1.6 1.4 -0.7 Absolute average annual change between 1995 and 2000 (percentage points): Imported materials (IM) % gross prod. 0.25 0.11 -0.06 IM from low-wage countries % production 0.1 0.04 -0.15 IM from high-wage countries % production 0.11 0.05 0.01 5.0 1.3 3.5 8.4 -28.9 55.6 3.9 -22.2 11.3 7.0 -27.3 55.6 0.43 0.23 0.15 0.7 -1.46 4.73 0.64 -1.96 4.84 0.18 -0.15 1.07 Data: Most important outsourcing sectors • Low-wage countries: – – – – – leather office machinery and computers TV, radio, communication equipment textiles, apparel basic metals • High-wage countries – – – – chemical products transport equipment and motor vehicles office machinery communication equipment Estimation results • • • • Negative and significant impact of imported materials from lowwage countries No impact of total imported materials Imported materials from high-wage countries are positive but not significant Sample split regressions: – – • negative and significant effect of total imported materials and imported materials from low-wage countries in less skill intensive manufacturing industries no effect in machinery, electrical, optical and transport equipment Quantile regressions – Effect of outsourcing is more pronounced at the low end of the conditional employment distribution (declining industries) OLS results, Labour demand (i) ln value added const. p. ln real wages imported materials (IM) % production (Q) IMQ low-wage countries IMQ high-wage countries constant 2 Adj. R # of obs (ii) (iii) (iv) coeff t coeff t coeff t 0.15 4.4 0.16 5.0 0.16 4.5 -0.31 -6.2 -0.32 -7.9 -0.34 -6.2 -0.07 -0.1 coeff t 0.17 4.9 -0.35 -5.9 -4.50 -3.3 -4.79 -3.5 0.68 1.2 0 -1.1 0.54 144 -0.01 -2.8 0.53 144 0.45 0.9 0 -1.1 -0.01 -2.9 0.54 0.50 144 144 Dep: var: average annual growth rate of total employment between 1995-2000. t-values are based on heteroscedasticity consistent standard errors. Empirical results OLS estimates weighted OLS estimates Median regression estimates OLS estimates weighted OLS estimates Median regression estimates constant imports from low-wage countries real wages value added c. p. Predicted employment actual employment Contribution of Sources of Labour Demand Growth (percentage points all manufacturing industries, total sample 0.53 -0.50 -0.77 -0.77 -0.49 -0.31 0.44 -0.21 -0.07 -0.07 -0.26 -0.04 0.42 -0.44 -0.41 -0.38 -0.28 -0.08 less skill intensive industries 0.31 -0.40 -1.30 -1.30 -0.74 -0.47 0.70 -0.35 -0.53 -0.53 -0.30 -0.58 -0.60 -0.59 0.32 -0.35 -0.25 -0.31 These calculations are based upon the average annual change in the explanatory variables multiplied by the regression coefficients. Conclusions • • • • Imports from low-wage countries have a negative and significant effect on employment Imports from industrialised countries have no effect Observed change in outsourcing accounts for an employment reduction of 0.26 percentage points per year. Magnitude of the effect differs across industries. – large effect in less-skill intensive industries – no effect in machinery, electrical, optical and transport equipment. – no effect in expanding industries • Future work: – Disaggregation of employment by skills => heterogenous labour demand – Outsourcing of services – Longer sample