1 / 40

LABOR TOPICS Nick Bloom Skill Biased Technical Change (SBTC)

LABOR TOPICS Nick Bloom Skill Biased Technical Change (SBTC). Why care about skill-biased technical change?.

lael
Download Presentation

LABOR TOPICS Nick Bloom Skill Biased Technical Change (SBTC)

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. LABOR TOPICSNick BloomSkill Biased Technical Change (SBTC)

  2. Why care about skill-biased technical change? It is a major topic in the literature – over 100 papers in the last two decades. There are a number of outstanding questions on this that careful micro-data work can addressKey political phenomena – Governments around the world have faced criticism that while their economic policies have increased the distribution of earnings (or employment).

  3. Changes in wage equality Skill Biased Technical Change (SBTC) Why this SBTC occurred

  4. Wage inequality over time Source: Autor, Katz and Kearney (2008, RESTAT)

  5. Wage inequality has been rising over time In the US wage (and consumption) inequality has risen since the 1960s Note the fall in female wage discount despite rising labor participation Source: Autor, Katz and Kearney (2008, RESTAT)

  6. What about by educational group: college/high school Source: Autor, Katz and Kearney (2008, RESTAT) Residual inequality is the variance of the error term (ei,t) from a Mincer wage equation: Log(wi,t) = α+βXi,t+ei,t

  7. This occurred throughout the period from 1960s Source: Autor, Katz and Kearney (2008, RESTAT) Note: The CPS data is available both from the NBER data section, and Census data from the Michigan IPUMS data site. Residual inequality is the variance of the error term (ei,t) from a Mincer wage equation: Log(wi,t) = α+βXi,t+ei,t

  8. This increase in inequality was particularly a phenomena of top half of the earnings distribution Source: Autor, Katz and Kearney (2008, RESTAT)

  9. Inequality also rising across educational groups Source: Autor, Katz and Kearney (2007, RESTAT) In a standard Mincerian regression the returns to a year of education rose from about 7.5% in 1980 to about 10% by 1995.

  10. At the same time the quantity of ‘skills’ has increased Source: Acemoglu (2002, JEL)

  11. The increase in skills happened both across and within industries Autor, Katz and Krueger (1998, QJE)

  12. The skills increase also happened within plants Source: Dunne, Haltiwanger and Troske (1997, Carnegie Rochester Conference Series )

  13. The international evidence • SBTC seems to have afflicted both global superpower nations • The UK experienced similar wage & employment trends as the US • Canada and Australia also experienced a similar phenomena • Across Europe there has been a more moderate wage experience – but typically more inequality in unemployment • This seems to be consistent with the idea that institutions constrained wages changes in Europe so movements in unemployment occur instead

  14. Changes in wage equality SBTC caused this change in inequality Why this SBTC occurred

  15. What has caused this within and between group changes in inequality? A summary response • Technology changes in much of the 20th century have been skill biased • This SBTC may have accelerated since the 1970s • The supply of skilled workers accelerated in the 1970s but slowed from the 1980s onwards • Thus, skills demand has outstripped supply, particularly since the 1980s, • raising between group (high/low education) inequality • The same phenomena has also probably also occurred for unmeasured • skills, raising within group inequality from 1970s onwards

  16. Why has technology been skilled biased (1/2)? • There is no need for technological changes to be skill biased • The industrial revolution in England increased the use of factories employing low skilled workers at the expense of craftsmen • Luddite rebellions of 1811 and 1812 were in response to falling wages of skilled weavers as factories replaced traditional weaving Ned Ludd – probably a fictional character but the movement was a major issue for the British, and even during the Napoleonic wars required extensive troops to surpress

  17. Why has technology been skilled biased (2/2)? • The support for 20th Century SBTC is empirical – there has been a • massive increase in the supply of skills (educated workers) at the same • time as skilled wages has risen, at least since 1970s. • This has happened in every sector of the economy – so a universal rise in • both the quantity and price of skills. This must be a demand shift • Evidence that SBTC driven earlier in the century due to electrification • (Goldin & Katz, 1998 & 2007)

  18. Over the 20th century skills premia has fluctuated • Source: Goldin & Katz (2007) • Variation in returns mainly due to change in relative supply of skilled and unskilled workers

  19. Also an interesting sharp-post war contraction in inequality – the “Great Compression” • Goldin and Margo (1992) argue arises because of: • Supply: Increased university enrollment (GI Bill), • Demand: Increase in non-skilled labor demand from manufacturing • Institutional: Unions strong post-war (low unemployment) and National War Labor Board

  20. Modeling the increase in returns to skills The traditional Solow model is skill neutral in technical change: Y=AKαLβHγ But the prior evidence suggests a strong skill biased component.

  21. Skill Biased Technical Change (SBTC) Can extend the Solow model to for skilled and unskilled labor L=[(AsLs)σ + (AuLu)σ]1/σ <1 SBTC in this setup would be the ratio As/Au rising over time Can substitute into a production function & re-arrange in terms of wage premium. Katz and Murphy (1992, QJE) did this and estimated the following regression implied by this production function: Ln(Ws/Wu)= β0 + β1(LC/LHS) + Dt + et They found β1≈-2/3 and Dt about 2.5% (2% on figures to 2005) Suggests that labor supply clearly matters, but there has been a steady trend favoring skilled labor over the last 40 years.

  22. Trends in college/high-school labor supply • Skill rose strongly in 1970s because: • Vietnam draft laws • Higher education expansion interacting with post-war baby boom • Can see 1970s rise in skills supply and falls in relative skilled wages against long-run trend Source: Acemoglu and Autor, (2010)

  23. Katz & Murphy (1992) results (updated by AA 2010) Once you detrend skills supply and relative wages the relationship is clear. Need to interpret cautiously, though, as only about 40 observations with serially correlated errors So predicted college/high school wage gap from a trend plus college/high-school skills supply looks a good fit But - need to interpret cautiously, as only about 40 observations with serially correlated errors Source: Autor, Katz & Kearney (2008, RESTAT)

  24. Changes in wage equality SBTC caused this change in inequality Why this SBTC occurred

  25. Why did this SBTC occur?

  26. Why did this SBTC occur - summary? • Proximate cause appears to be cheaper capital and/or computers • But why is this skill-biased? Several arguments: • Skills directly complement capital • Skills directly complements computers • Skills needed for rapid change – post 1970s had rapid change • Other factors that appear to play an additional (more minor) role: • Labor market institutions (minimum wage and Unions) • Trade with developing countries, e.g. China • But why did capital (particularly PCs) become cheaper? One view is the direction of technology is endogenous – the rise in skills promoted SBTC to occur

  27. (a) Capital complementarity (1/2) One plausible idea is that capital is more complementary to skilled labor then unskilled labor. Krussell, Ohanian, Rios-Rull and Violante (2000, Econometrica) Y=Kα(λ[μKsρ + (1-μ)Lsρ ]σ/ρ + (1- λ )Luσ)1/σ If σ>ρ then reductions in the cost of K increase the demand for Ls Effectively this replaces As/Au with the price of capital

  28. (a) Capital complementarity (2/2) Krussell et al. (2000) then provide evidence for a long-run fall in the cost of capital providing results for the model matching the data So neat model and plausible results. But there is an identification problem as the impact of the cost of capital is killed by a time trend (Acemoglu (2002, JEL), so can not be certain.

  29. (b) Computer capital complementarity (1/3) Worker-level evidence Krueger (1993) shows that people using computers earn higher wages, and this wage premium has increased over time. Consistent with computers playing an important role, but also with computers proxying unobserved skills – for example DiNardo and Pischke (1997) show similar phenomena is true for pencils. R&D also correlated ≈0.8 with computer use Machin & Van Reenen (1998) Computers or pencils?

  30. (b) Computer capital complementarity (2/3) • Industry level evidence • A number of papers also show that: • All industries show an increase in skills demand and skill premium • This rise is faster in industries increasing computerization faster • The drawback to this evidence is that: • Unobserved – could have been something else driving both • Increase in computerization in the 1980s also predicts skills premium increases in the 1960s • In summary, appears likely computerization is strongly linked with SBTC, but hard to prove definitively

  31. (b) Computer capital complementarity (3/3) • Most recently Autor, Levy and Murnrane (2003) use the Dictionary of Occupational Titles to allocate cognitive and manual repetitive and non-repetitive tasks to jobs • Idea is repetitive tasks can be replaced by computers, non-repetitive ones can not • Find that wages and employment in repetitive tasks fallen fastest – leading to a polarization of employment: “lovely and lousy jobs” as christened by Goos and Manning (2008) for the UK

  32. The polarization of employment (US data) Evidence that employment is polarizing since the early 1990s – employment growth strongest below 30th percentile above the 75th Source: Autor, Katz and Kearney (2007, RESTAT)

  33. The polarization of employment (International data) Source: Acemoglu and Autor (2010, HLE)

  34. The polarization by occupation (US data) Source: Acemoglu and Autor (2010, HLE)

  35. (c) Skills are needed to deal with change • The Nelson and Phelps (1966) hypothesis is that change is complex and skilled people are better at dealing with this • The “acceleration hypothesis” • Consistent with evidence that higher skilled employees are increasingly in demand as firms rapidly changing technologies • Problems are that periods of 1970 to 1995 are associated with sluggish TFP growth – hard to reconcile this with radical technological change • So in summary seems plausible but hard to fully pin down

  36. What about other factors – trade unions? • Trade unions have been weakening since the 1970s while the absolute level of the minimum wage fell strongly in the 1980s • This almost certainly played a role in the particularly poor performance of the lower earnings quartiles in the 1980s. • But: • unions weakened only in the 1980s while the changes in inequality started in the 1970s • unions only likely to effect lower quartiles, while higher quartiles is where most of the action was

  37. What about other factors – minimum wage? • Real value of the minimum wage fell throughout the 1980s as this was not indexed and frequently not updated. This almost certainly played a role in particularly poor performance on the lowest quartile in the 1980s. • But problems with MW as a complete story: • MW only started to decline in real-value in 1980s • Other countries – like the UK – had no MW until late 1990s

  38. What about other factors – international trade(1/2) ? • Trade from China, India and other countries could also play a role? • Literature generally discounts this as a major force as skill levels and high-skilled wages have risen in almost every industry (including all the non-tradable sectors) • Also trade generally has limited predictive power: e.g. Berman, Bound and Griliches (1994, QJE), Autor, Katz and Krueger (1997, QJE) and Machin and Van Reenen (1998 QJE) • The basic Hecksher-Olin model would not predict this within industry effects (predicts between industry effects going against within industry)

  39. What about other factors – international trade(2/2) ? • But: • Could possibly be due to outsourcing within non-tradable industries • More generally the empirical evidence is primarily in late 1990s before Chinese imports really took off. Since then entire industries have virtually disappeared (furniture, toys, textiles etc..) • So trade is probably an increasingly big factor: Bloom, Draca and Van Reenen (2011) finding major effects only post 2000 (particularly 2005)

  40. Endogeneous technical change Final question is why did SBTC occur in the 1970s?Acemoglu (1998, QJE and 2002 RESTUD) and others have a number of papers around the idea of endogenous technical change – idea that increased supply of graduates led to technical change Related idea is endogenous technical adoption – different countries adopt different technologies endogenously An interesting area of research and plausible hypothesis but with limited empirical evidence beyond particular examples like drugs (Acemoglu and Linn, 2004 QJE) and air-conditioners (Newell, Jaffe and Stavins, 1999 QJE)

More Related