User talk:Wiki Presentation - Topic 11

The Technological Change Explanation for Increased Wage Inequality
Technological change has undeniably been prevalent and influential over the last few decades, having an impact on every sector of society both in the developing world and also the lesser developed world. It is arguable whether these advances have had a positive effect on global employment or whether it has simply exaggerated the already widening wage differential between skilled and unskilled workers.

Contents:
 * 1. SBTC
 * 1.1  SBTC explanation
 * 1.2  Criticisms
 * 2. Measuring inequality in wages within skill groups and the extended model
 * 3. Polarization
 * 3.1  Routinization
 * 3.2  The model
 * 4. Alternative Concepts
 * 5. References

1. SBTC

1.1 The skill-biased technical change explanation SBTC strives to find an explanation to the complementary nature of the simultaneous supply and wage increase and of how an increase in skilled labour can consequently result in a wage increase. Under the hypothesis that increasing skilled labour supply can create an incentive for investors to create skill complementary technology, then those innovations would increase the productivity of the skilled workers. This in turn would increase their earnings.

It states that technological innovations complement high-skilled labour, rather than acting as worker replacement. Technology always increases the productivity of labour, but because of its unique complementing nature, it does not replace jobs. Therefore, this increase in productivity leads inevitably to a wage increase. A large supply of skilled labour provides profit incentive to create technologies that can be utilized by the supply.

Labour market composition changes. We will have people earning more or less depending on the job they have, and how important it is depending on the market needs. A labour market demand composition change would thereby alter the wages earned by any single category. The more educated, assumed to have higher-skill, receive the highest wages and also had the highest rate of income growth over the period. Thus, the more educated become more productive, specially relating them with their less educated counterpart. Because of this, employment increases for the most educated, high-skilled workers. Evidence shows, that both the supply of skilled labour and the wages earned by skilled workers increased.

1.2 Criticisms
 * Questions the assumption that skilled labour is defined by educational level; This hypothesis should not be based on education being a solely causal factor and that it may be that occupation is what leads individuals to be productive. Proclaims that a change in the labour composition of occupation played a role in wage dispersion and not just the educational level; There is a large growth of female professionals in the work force (labour composition change). Earnings trend reflects a change from less productive to more productive work. We can even say that there is a reallocation of labour power.


 * Cannot explain the dispersion by itself.

''
 * Globalization; Another factor that should be considered is globalization. It acts to further increase the productivity of already productive labour. A highly productive good or service will only become more demanded as global markets integrate. Therefore, technology that further increases the marginal revenue of labour becomes additionally profitable.


 * Are the large measured wage differentials associated with on-the-job computer use productivity gains or the result of unobserved heterogeneity?


 * ''The evidence supporting a major role for skilled-biased technical change, however, is primarily inferential.

 Until recently, changes in the wage structure have rarely been directly linked to observable technical developments that are presumably transforming workplaces and the distribution of wages.

Whether the measured returns to computer use reflect productivity differences due to the introduction of computers in the workplace or whether these returns merely reflect unobserved heterogeneity. If the latter is correct, the direct evidence that skill-biased technical change explains changes in the labour market is severely undermined.

We find that when evaluating the tools used at work that the measured return to the use of pencils at work is almost as large and robust as the measured return for computers use. Since we do not believe that pencils change the wage structure, this would appear to undermine the view that the coefficient on computers provides direct evidence on the role of skill-biased technological change.

Private returns only increase to those workers who possess a skill like the ability to programme a computer or work with business software. If these workers are scarce, their wages should be bid up in the market. In addition, the return should increase to everybody possessing the skill regardless of whether the worker actually uses a computer or not. Therefore the correct variable to be calculated should be computer knowledge rather than computer use.

Also, it only makes sense to employ a worker with computer knowledge in a non-computer job if this worker is at least as productive in the alternative job, suggesting that the wage differential could be credited to some other skill or attribute that is often found in the same workers who possess computer skills.

The returns to school tend to be slightly lower at very low educational levels. Workers who have not completed elementary school, very rarely use computers. However, this is US data. When we interact computer use and education in German data we find the opposite result.

Dinardo states: a complementary story for rising inequality focuses on the decline of unions, the real value of the minimum wage, and pay-setting norms that have historically served to compress the wage structure.1,3,5

2. Measuring Inequality in wages within skill groups and the extended model

The original (canonical) model of skill-biased technological change (SBTC) states that the total wage each worker receives depends on his efficiency unit and unit wage. Usually, a worker is classified into either a high-skill group or a low-skill group, empirically, in accordance with his education.

If a worker belongs to a high-skill group (or a low-skill group), his unit wage is $$M$$$h$ (or $$M$$$l$ respectively). Thus, a total wage for a high-skill worker $$i$$ ∈ $$h$$ is $$W$$$i$ = $$M$$$h$×$$Y$$$i$ where $$Y$$$i$ denotes efficiency units by the high-skill worker $$i$$ ∈ $$h$$. In the simple form, inequality in wages within group can be stated in terms of the relative earnings of two workers in the same group. For example, if the two workers are in a high-skill group, i.e. $$i,j$$ ∈ $$h$$, mathematically their relative earnings are defined as $$W$_{$i$}/$W$_{$j$}$ = ($$M$$$h$×$$Y$$$i$) ÷ ($$M$$$h$×$$Y$$$j$) = $$Y$_{$i$}/$Y$_{$j$}$. However, from this equation, the inequality is independent of the actual skill premium.

To incorporate the idea of actual skill bias (which now cannot be approximated by education or experience), the simple model of the SBTC has to be extended by drawing a distinction between observable groups (such as college vs. non-college) and skills.

Assume that the two observable groups are college and non-college, and a fraction $$G$$$c$ of college graduates are highly skilled, while a fraction $$G$$$n$ < $$G$$$c$ of non-college graduates are highly skilled (the remaining fractions of both groups, $$1-G$$$c$ and $$1-G$$$n$, are low skilled). Also, denote the actual skill premium by $$M$$ = $$M$_{$h$}/$M$_{$l$}$, which is not a college premium anymore as skills now cannot be approximated by education. In other words, not all college workers have high skill, and not all non-college workers have low skill. From these assumptions, the college premium, $$M$$$e$, is the ratio of average college wages, $$M$$$c$, to average non-college wages,$$M$$$n$:

$$M$$$e$ = $$M$_{$c$}/$M$_{$n$}$ = [$$G$$$c$$$M$$$h$ + ($$1-G$$$c$)$$M$$$l$] ÷ [$$G$$$n$$$M$$$h$ + ($$1-G$$$n$)$$M$$$l$] = [$$G$$$c$$$M$$ + ($$1-G$$$c$)] ÷ [$$G$$$n$$$M$$ + ($$1-G$$$n$)].

Obviously, since $$G$$$n$ < $$G$$$c$, this college premium, $$M$$$e$, is increasing in $$M$$. In other words, if the price of skill rises, then the college premium also rises. Additionally, within group inequality is defined in this context as the ratio of the earnings of high-wage college graduates and non-college graduates to those of low-wage college graduates and non-college graduates. Hence, $$M$$within = $$M$$, as high-wage workers in both groups earn $$M$$$h$ while low-wage workers earn $$M$$$l$. As long as $$G$$$c$ and $$G$$$n$ remain constant, $$M$$e and $$M$$within will move together.

This extended model predicts that there is a relationship between an increase in the returns to observed skills, such as education, an increase in the returns to unobserved skills. Moreover, some manipulations of compositional changes can be made, affecting $$G$$$c$ and $$G$$$n$, so within group inequality can change differently than the skill premium. Thus, overall inequality can exhibit more complex changes as supplies and technology evolve.

The study also reveals that the increase in the overall earnings inequality starting in the late 1970s or early 1980s was associated to the increase in the demand for skills, which reflected in the increase in the college premium. Overall, this model is capable for analyzing the evolution of distribution of earnings; nonetheless, it does not give sufficient understanding of why different parts of the earnings distribution, moving differently in different time periods.

3. Polarisation of the Labour Market

There has been a noted increase in both wage polarisation (income shifting to high and low wages) and job polarisation (growth in the share of employment of both high-skill high-wage jobs and low-skill low-wage jobs). As we can see in Figure 3, although there is an increase in the percentage change in employment share of high-skill high-wage occupations predicted by SBTC theory there is also, contradictory to SBTC theory, an increase in the percentage change in emplorment share of low-skill low-wage workers. The main theory as to why this happens is the theory of routinization set forward by David Autor and his co-authors.1,6

Alternate theories such as the increased feminization of employment (more women in the labour force typically working in part-time and therefore low paid jobs), and a general increase in part-time low-paid jobs are disproved as there are similar trends for both males and females, and whether the data takes into account total hours or not.

3.1 Routinization

Firstly, it is important to make the distinction between five different types of task that Autor, Levy and Murnane put forward. Those are non-routine cognitive, non-routine interactive, routine cognitive, routine manual and non-routine manual. These tasks could then be associated to different occupations using the US Dictionary of Occupational Titles (see Table I). These can also be classified as abstract labour, manual labour and routine labour. Abstract labour (problem solving, intuitive roles) is monotone up education level and normally associated with high-skill high-wage roles, manual labour (involving human interaction and environmental adaptability) is monotone down education level and normally associated with low-skill low-wage roles and routine is non-monotone in education.1,6,7

The increase in productivity and decrease in real price of information and computer technology, especially over the last twenty years, is thought to have had the effect of routinization. This is the increased substitution of middle-skill middle-wage labour for technology; the computerization of routine manual and cognitive tasks to decrease costs. Routine tasks such as clerical work, administration and some production line work can be ‘codified’ for and replaced by technology. However, a computers ability to make judgements and respond to its environment is entirely reliant on whether a programmer can write a set of rules for the computer to follow. Therefore computerisation could be considered to be complimentary to abstract roles, such as programmers, exacerbating the polarization already caused by routinization.1,4

3.2 The Model 𝑌=𝐴𝛼𝑅𝛽𝑀𝛾 𝛼,𝛽,𝛾 𝜖 (0,1)  &   𝛼+𝛽+𝛾=1 In this model Y represents total output, A represents abstract tasks, R represents routine tasks and M represents manual tasks. We consider these to be roughly correlated to high-skill, middle-skill and low-skill workers respectively. We assume there are two types of workers: those that have gone to college and perform abstract tasks and those that have completed high school and substitute between routine and manual tasks. We also note here that computer capital is labelled ‘K’ and is measured in efficiency units. It is a perfect substitute for routine labour (represented by LR) and is supplied perfectly elastically at price ‘p’ which is falling at a constant exogenous rate (as technology advances

the price of it is expected to decrease – compare the real price of a computer twenty years ago to now). LA and LM represent the labour supplied by abstract and manual workers.

Each college worker is assumed to have one efficiency unit of abstract skill which they supply to abstract tasks inelastically. Each high school worker (i) is assumed to have one efficiency unit of manual skill, ηi units of routine skill with η ϵ [0,1] and no abstract skill. Each high school worker (i) decides whether to pursue manual or routine tasks. This decision will depend on wages wm for manual tasks and wr for routine tasks. If ηi<wm/wr then high school workers will supply one efficiency unit to manual tasks otherwise they will supply ηi efficiency units to routine tasks. Therefore the labour supply of manual tasks is weakly upward sloping in wm/wr. Because computer capital, K, is a perfect substitute for routine labour the price of computer capital is equal to the price of routine workers, ie wr=p.

We reach equilibrium when supply equals demand, each worker is paid their marginal product and the labour market clears so that workers wish to reallocate themselves to different tasks.

Now we have looked at the whole model we want to see what happens during technological change ie. There is a decrease in p. A decrease in p leads to a one-to-one decrease in wr. This results in:


 * 1) A decline in p leads to an increase in the demand for routine tasks
 * 2) Due to workers self-selection extra demand is supplied by computer capital, K. A rise in routine input also raises the marginal productivity of manual task output – wm increases. As p decreases and therefore so does wr some high-school workers, especially those with low routine efficiency, switch their allocation to manual tasks.
 * 3) This additional manual labour supply works against the benefits of computerisation on manual wage. It is therefore possible for both wr and wm to fall. However, the wage of manual relative to routine (wm/wr) will always rise.
 * 4) Abstract workers also benefit from increased routine output due to having complimentary tasks. As workers cannot reallocate themselves to abstract tasks the positive effects of computerisation on wA is not counteracted by an increase in the supply of workers.4
 * 1) This additional manual labour supply works against the benefits of computerisation on manual wage. It is therefore possible for both wr and wm to fall. However, the wage of manual relative to routine (wm/wr) will always rise.
 * 2) Abstract workers also benefit from increased routine output due to having complimentary tasks. As workers cannot reallocate themselves to abstract tasks the positive effects of computerisation on wA is not counteracted by an increase in the supply of workers.4
 * 1) Abstract workers also benefit from increased routine output due to having complimentary tasks. As workers cannot reallocate themselves to abstract tasks the positive effects of computerisation on wA is not counteracted by an increase in the supply of workers.4

4. Alternative Concepts

The key problem that the SBTC theory faces is the arbitrary measurement of technological change. The majority of regression models use transcendental logarithmic (translog) equations as suggested by Christensen, Jorgenson and Lau (1973), these give relatively accurate and simple results however often face problems when econometric analysis is applied and cause issues when forecasting. These problems occur in the following forms:
 * 1)  It is hard to differentiate the fixed effects of technological change
 * 2)  There is often endogeneity between variables
 * 3)  Measurement problems in discerning the units of a technology based variable
 * 1)  There is often endogeneity between variables
 * 2)  Measurement problems in discerning the units of a technology based variable
 * 1)  Measurement problems in discerning the units of a technology based variable

The measurement of technology is the gravest problem that affects regression analysis on SBTC. For example, a key leading indicator for improvement in technological change is that of Research & Development funding. However, due to the fact that Research and Development funding’s effect on actual output is often difficult to measure as an economist, it is very hard to determine whether a particular firms knowledge base is changing over time. To counteract these problems, a number of different methods have been practiced over time, the most practiced are the following:

4.1 SBIT - Skill-Biased International Trade

This theory suggests that increasing economic integration with less developed labour intensive countries induces a wage reduction of low skill workers in the more advanced countries. A similar situation to what we have seen in the EU where increased integration produces a platform that supports migration. Many economists measure skill-biased international trade by calculating the skilled to unskilled ratio of a country or countries with the ratio of non-production-to-production workers. As these are tangible assets that are easy to measure make it a more accurate tool in measuring fixed affects.

4.2 The Supply of Human Capital

This approach puts emphasis on the supply of labour – in particularly the rise in the Global Supply of skilled workers due to better educational opportunities for minorities and support for those who previously had been neglected by the educational system. Emphasis is also put on the composition of labour supply; stating that a change in the composition could strengthen or weaken the demand for labour shifts. The Supply of human capital is more accessibly measured than SBTC by using such factors as; percentage of population who complete A-levels or differential in people who have a degree.

4.3 LMI – Labour Market Information

LMI focuses on institutional factors that could influence wage differentials such as de-unionisation, reductions in the minimum wage and labour law. These factors are often easier to measure than technological change, but due to the dependence on data from Governments that are not always available at regular intervals, and this theory is often rejected.

5. References
 * 1) Daren Acemoglu and David Autor (2011), “Skills, Tasks and Technologies: Implications and Earnings” Handbook of Labour Economics, 4, 1043-1171
 * 2) David Autor, Lawrence Katz and Alan Krueger (1998), “Computing Inequality: Have Computers Changed the Labour Market?”, Quarterly Journal of Economics 113
 * 3) John DiNardo and Steffen Pischke (1997), “The Returns to Computer Use Revisited: Have Pencils Changed the Wage Structure Too?”, Quarterly Journal of Economics 112
 * 4) David Autor, Lawrence Katz and Melissa Kearney (2006), “The Polarization of the US Labour Market", The American Economic Review, 96, 189{194.
 * 5) http://marginalrevolution.com/marginalrevolution/2006/12/does_skillbased.html
 * 6) Maarten Goos and Alan Manning (2007) “Lousy and Lovely Jobs: the Rising Polarization of Work in Britain” The Review of Economics and Statistics 89
 * 7) David Autor, Frank Levy and Richard Murnane (2003) “The Skill Content of Recent Technological Change: An Empirical Exploration”, The Quarterly Journal of Economics
 * 8) Jaewon Jung and Jean Mercenier (2010) — Preceding unsigned comment added by 86.7.40.215 (talk) 15:58, 25 November 2012 (UTC)
 * 1) http://marginalrevolution.com/marginalrevolution/2006/12/does_skillbased.html
 * 2) Maarten Goos and Alan Manning (2007) “Lousy and Lovely Jobs: the Rising Polarization of Work in Britain” The Review of Economics and Statistics 89
 * 3) David Autor, Frank Levy and Richard Murnane (2003) “The Skill Content of Recent Technological Change: An Empirical Exploration”, The Quarterly Journal of Economics
 * 4) Jaewon Jung and Jean Mercenier (2010) — Preceding unsigned comment added by 86.7.40.215 (talk) 15:58, 25 November 2012 (UTC)
 * 1) David Autor, Frank Levy and Richard Murnane (2003) “The Skill Content of Recent Technological Change: An Empirical Exploration”, The Quarterly Journal of Economics
 * 2) Jaewon Jung and Jean Mercenier (2010) — Preceding unsigned comment added by 86.7.40.215 (talk) 15:58, 25 November 2012 (UTC)
 * 1) Jaewon Jung and Jean Mercenier (2010) — Preceding unsigned comment added by 86.7.40.215 (talk) 15:58, 25 November 2012 (UTC)