Topic 4 -Technological change, employment and wages Professor Christine Greenhalgh P Cahuc and A Zylberberg (2004) Labor Economics, Chapter 10: Technological Progress, Globalization and Inequalities, parts 2 and 3. C Greenhalgh and M Rogers (2010) Innovation, Intellectual Property and Economic Growth, Chapter 10: Technology, Wages and Jobs, Princeton University Press. Hornstein, A., P. Krusell and G. L. Violante (2005), 'The effects of technical change on labor market inequalities', in Handbook of Economic Growth, Volume 1B, P. Aghion and S. Durlauf, (eds.), Amsterdam: North Holland/Elsevier B.V. Does new technology destroy jobs? Two kinds of innovation with different impacts: Process innovation – new ways of making and delivering products Effects of Process Innovation: • New technique increases efficiency and thus lowers costs of production • Fewer workers can produce same output • This can cause technological redundancy BUT • Cost reduction may lead firm to expand its output as it gains market share • Potentially this leads to more jobs on balance Innovation to create demand Product innovation – firm brings new varieties and qualities of products to the market Effects of Product Innovation: • Firm can capture new or increased segments of markets • Again this is likely to lead to more jobs Why the fear of new technology among workers? This is a longstanding issue: • Luddites (early 19th century England) smashed new equipment being installed in textile industry • Saw this as destroying their craft jobs and permitting unskilled labour to take over their work at lower wages Labour augmenting technological progress with two factors If assume a CES production function and cost minimisation in production of given output level then it can be shown: Demand for labour depends on level of output (Y), real wage (W/P), degree of substitutability (σ) between capital (K) and labour (L) the rate of labour augmenting technological progress (A) ln L = ln Y – σ ln W/P + (σ – 1) ln A Source: Van Reenen (1997) J. Lab Econ. Summary in Greenhalgh and Rogers Ch 10 What happens to demand for labour as its efficiency improves? From the above model it can be shown that elasticity of labour demand w.r.t. labour augmenting technol. change (ΔA) is: ηLA = ηP θ + (σ – 1) where (σ – 1) is the ‘substitution effect’ of ΔA: +σ use more L as now more cost effective – 1 as get more output per worker by ΔA and ηP θ is the ‘scale effect’ of expanding Y: ηP is price elasticity of output demand θ is production cost reduction effect of ΔA Effects of ΔA (labour augmenting) Good news for workers (ηLA is +ve) if: • Capital and labour are easily substituted (σ is large) • Cost savings are passed through to customers (θ is significant) • Product demand is price elastic (ηP is large) Bad news for workers (ηLA is -ve) if: • Product has highly inelastic demand (ηP small) • Cost savings are kept in firm to raise profits (θ = 0) • There is very little substitutability between capital and labour (σ is small) Employment growth and innovation in firms in Europe 1998-2000 Source: Table 10.1 of Greenhalgh and Rogers drawn from Harrison et al. NBER WP 14216 (2008) France Germany Spain UK Manufacturing employment growth 8.3 5.9 14.2 6.7 Process innovation - 0.1 - 0.6 0.3 - 0.4 Product innovation 5.5 8.0 7.4 4.8 Services employment growth 15.5 10.2 25.9 16.1 Process innovation - 0.1 0.1 0.0 0.2 Product innovation 8.0 7.6 6.5 5.4 Features of this data from Community Innovation Survey • Expansion of services employment approx. twice that in manufacturing firms over two year period • Consistency across four EU countries in relative size of product and process innovation effects • Firms introducing new products attributed a significant share of their employment growth to this activity • Process innovation was broadly neutral, with biggest negative effect being 0.6% in Germany • Do these findings lay the Luddites to rest? Time series empirical evidence about innovation and employment Greenhalgh et al. Scottish JPE, V 48(3) 2001 Data – panel of UK industrial firms 1987-94 Estimated demand for labour function including measures of their innovative activities • Doing R&D associated with raised employment • Predicted effect of doubling R&D (= German levels) in R&D active firms -> 2.6% rise in jobs • Gaining patents also gives higher employment • One more patent per firm -> 0.4% rise in jobs • Dividing sample by high-tech and low-tech: – R&D impact larger (4%) in high-tech sectors – patent impact larger (1%) in low-tech sectors Innovation and wages in firms – micro aspects • Rent sharing with innovation – Innovation raises profits and affords some monopoly power to firm – Firm shares some of returns to raise worker loyalty (efficiency wage argument) • New processes embodied in better machinery, computers and robotics – Increased productivity for complementary workers raises their wages (designers, programmers, managers, technicians) – Reduced demand for substituted workers causes lowering of their wages (shop floor workers, call centre workers) Innovation and wages - evidence • Van Reenen (1996) data for GB 1976-82 showed innovation led to rises in profits and rent sharing occurred as 20-30% awarded to workers in wage rises • Greenhalgh et al. (2001) data for UK 1986-95 found positive effect on wages both when firm is doing R&D and when making use of trademarks (indicator of product launch) • Krueger (1993) US data for 1980s, estimates that workers using computers earn a premium of 10 – 15% • Entorf and Kramarz (1997) for France caution that those selected to work with computers are the more able, so wage gain is more modest Unions and innovation - theory • Are unions a negative force? If they raise wages and resist changes to work practices does this reduce firm’s incentive to innovate? • Interaction between product markets and structure of bargaining makes a difference to their impact (Dowrick and Spencer) • Resistance to new technology is more likely if craft or industry union and oligopoly product market • Resistance is less likely if enterprise union and competitive product market • Are unions a positive force? (Freeman and Medoff) If unions improve workers’ job satisfaction and improve communications, then turnover falls, training improves, perhaps innovation more likely? Unions and innovation – evidence (Summarised in Greenhalgh and Rogers Ch 10) • Evidence for US (Hirsch) indicates unions have a negative effect on both capital investment and level of R&D • Evidence for UK (Menezes Filho et al.) says this simple correlation is due to most unionised firms being in declining industries • Results for UK and US differ because unions in UK want to protect jobs but in US want to raise wages • Evidence for UK that union presence raises vocational training (Booth et al.) • Evidence for UK that industry skill shortages are associated with lower investment and reduced R&D (Nickell and Nicolitsas) Innovation, jobs and wages - the macro picture Source: Table 10.1 of Greenhalgh and Rogers drawn from Machin (2001) Ox Bull Ec & Stats Vol 63 Special Issue Share of graduates in total employment (%) Relative wages of graduates to non-graduates US UK US UK 1980 19.3 5.0 1.36 1.48 1990 23.8 10.2 1.55 1.60 2000 27.5 17.2 1.66 1.64 Change in ratio of earnings at the median to bottom decile (D5/D1) Source: Cahuc and Zylberberg Table 10.2 Late 1970s Mid 1990s Change France 1.68 1.60 – 0.08 Germany 1.52 1.46 – 0.06 Japan 1.58 1.60 0.02 UK 1.58 1.80 0.22 US 1.93 2.20 0.27 Reasons for the shift in demand towards the skilled workers In remainder of the lecture we compare three possible sources of skill shift in demand for labour in rich countries: • Skill-biased technological change • Globalisation and specialisation in trade • Changes in composition of final demand Perhaps all three have operated at once? Relative wages, differential productivity and supply growth (Source: Katz and Murphy 1992, in Hornstein 2005) Assume two types of labour, skilled and unskilled with wages ws and wu respectively Elasticity of substitution between labour types is σs Relative wage of skilled to unskilled labour is driven by two ratios: • Difference in productivity growth of each type of labour, As / Au • Relative supply of each type of labour, ls / lu ws s 1 As 1 ls ln ln ln wu s Au s lu Predictions of simple model for relative wage of skilled/unskilled • If productivity of skilled labour rises faster than that of unskilled labour, relative wage for skilled workers will rise • If supply of skilled labour rises faster than that of unskilled, then relative wage will fall • The higher the degree of substitutability between skilled and unskilled (value of s) then – the larger is the positive effect of rising relative productivity on relative wages – the smaller is the negative effect of rising relative supply Model with two types of labour and capital equipment Source: Hornstein, or see Greenhalgh and Rogers Elasticity of substitution between unskilled labour and equipment is ue = 1/(1-), and for skilled labour and equipment is se = 1/(1-) Unskilled labour is more easily substitutable with equipment than is skilled labour ue > se ( > ) The relative wage equation now has three elements: ws ln wu ue 1 As 1 ls ( ) ke ln ln ls ue Au ue lu Predictions of three-input model for relative wages of skilled/unskilled • Added effect driving demand for skilled labour is it is complementary with capital equipment • Relative wage of skilled workers rises with any increase in ratio of equipment to skilled labour • Innovation has improved productivity of capital so an increase in capital intensity has occurred • Big rise in computer use, especially in services sector, has increased demand for skilled labour • In manufacturing the use of robots and other automation has reduced demand for unskilled • Evidence for US - these factors explain much of change in relative wages from 1960s to 1990s Globalization - Is international trade also skill biased? • Asian development 1970s & 80s ‘the Asian tigers’ (Hong Kong, Singapore, S. Korea Taiwan) - made small inroads into Western manufacturing • More Asian development 1990s (China and India) jointly have 37% of world population) so have much larger impact on world trade • HOS model of trade based on domestic factor endowments predicts specialisation by factors • Opening up of countries with large supply of low cost unskilled labour leads rich countries to specialise in goods using skilled labour • Employment and wages of unskilled labour in West expected to fall (see Wood 1994) Demand - A third cause of skill bias? • Income growth in rich countries has been steady and sustained over last 25 years • Composition of demand will change due to varying income elasticity of demands for goods and services • Luxuries (income elastic) account for more spending than necessities and demand for inferior goods falls as incomes rise • High technology innovative products require skilled labour to design and produce and • Relative demand for these will grow as these innovative products will be in luxury category Three causes of skill bias in demand for labour, UK 1979-90 Source: Greenhalgh and Rogers Table 10.3, from Gregory et al. Oxford Economic Papers, 2001 Total % change in employment Final demand Net exports Technological change High skill 28.8 28.2 – 4.1 4.6 Intermediate skill 0.1 21.1 – 4.8 – 16.2 – 14.9 17.9 – 5.7 – 27.1 3.5 22.0 – 4.8 – 13.7 Low skill Total change Implications of these empirics • Calculations used Leontief Input-Output model with linear production technology • This attributes all output growth to the category ‘change in final demand’ even if some is due to product innovation • ‘Technological change’ effect is measured by factor inputs needed to produce a constant output • Changing technology in production gave rise to biggest relative demand shift to skills (30% twist) • Growth in final demand added to relative demand shift (10% difference in growth) • Trade shows rather modest effects (but database pre-dates rapid growth of China) • Technology changes may be stimulated by trade