Presentation on theme: "1. Shape of the earnings/wage distribution 2. Measures of dispersion 3. Geographical dispersion (country/region) 4. Dispersion according to education/skills."— Presentation transcript:
1. Shape of the earnings/wage distribution 2. Measures of dispersion 3. Geographical dispersion (country/region) 4. Dispersion according to education/skills 5. Age 6. Gender & ethnicity which will be discussed after chapters 9 & 10) 7. Family (marital status, children) 8. Sector, firm and industry 9. Long-term vs short term distribution 10. Intergenerational mmm 11. Change over time (US, Sweden)df s erar
Distribution of wages, earnings and income tend to be log-normal. The logarithm of the wage is normally distributed. The distribution of the wage is skewed to to the right (has a long ”tail” on the right side.) Wage dispersion has two elements: Difference in skills between workers Difference in wages between workers with the same skills. The level and form of dispersion depends on both market factors and institutional factors.
Overall measures ◦ (Coefficient of) variation (std. dev./mean) ◦ Gini coefficient Gini koefficient is derived from the LORENZ CURVE The Lorenz curve is based on the distribution over percentiles n percent of the population have income/wages below the n th percentile ◦ Percentiles, deciles, quartiles, quintiles…
List all individuals in order of increasing wage. Number one is the person with the lowest wage in the population, number two has the second lowest and the person with the highest wage is last on the list. Arrange them in this order along the x- axis. On the y-axis, mark the cumulative percentage of all wages that accrue to the people”to the left” of this person.
Assume that there are 10 wage earners: Personwage cumulative share A44 B610 C717 D825 E934 F943 G1053 H1265 I1580 J20100
Area II Area I Area I+Area II = 0.5 Area I/(Area I+Area II) = 2*Area I = the GINI COEFFICIENT
Different distributions can have the same Gini-coefficient. Two alternative measures: ◦ The share of the highest 10 percent/the share of the lowest 10 percent ◦ P90/P10 (Why are these two unequal? Which is largest?) To see whether there is most inequality at the upper or lower end of the distribution one can use: ◦ P90/P50 and P50/P10 (or P75/P50 and P50/P25) ◦ The shares of the lowest and highest 10 percent.
TotalWomenMen P P50 median P P90/P108,07,38,1 Gini-koefficient 0,3480,3220,350
In the beginning of industrialisation, wage and income disparities tend to get very large. As industrialisation progresses, historically, inequality has decreased. Richer countries tend to have smaller income inequality than poorer countries. But there are large differences between OECD countries and larges differences between developing countries.
Source: OECD, Employment Outlook 2010 Earnings dispersion, some OECD countries 2010
RegionWomen and men Mean (SEK)P10MedianP90 P90/ P10 P90/ P50 P50/ P10Gini SWEDEN ,201,853,350,353 Stockholm ,842,085,200,404 Karlstad ,901,794,420,345 Danderyd ,622,835,170,503 Bjurholm ,641,762,060,279 Source: Statistics Sweden,
In the public sector compared to the private: Less dispersion Smaller gender differentials Earlier higher wages for both men and women, both unadjusted and adjusted for education (Level of Living Surveys ) Now lower average wages than private sector (LLS 2001) The difference is largest in the highest deciles.
We have seen that wages increase with education Source: SCB Statistikdatabasen
UniversitySec./ /secondaryprim Adjusted for age and gender Source: LNU (from Björklund et. al.)
Returns to secondary education ◦ dropped sharply from 1968 to 1974 ◦ have varied a little but no distinct trend afterwards Returns to university education: ◦ Gradual (and large) decline from 1968 to some point in the 1980s ◦ From mid-1980s new increase but not to the levels of the 1960s
Another way of measuring: The returns per year of schooling estimated in wage equations (controlling for experience, industry, sector etc.) each year There is a steady increase, from 5 to 6 percentage points for men, 3 ½ - 4 ½ for women. For education which is at the right level for the job returns are somewhat higher. Source: Johansson and Katz (2007)
In the US (see Borjas fig. 7-5) the wage differential between college and high school graduates decreased in the 1970s but increased from about 50 percent to 90 percent from 1980 to The wage differential between those with and without full high school education also increased to but less dramatically.
Demand: We think that there has been a general increase in demand for highly educated workers but it is hard to measure. Supply: There was a large increase of workers with long university schooling in the 1970s and first half of the 1980s but slower afterwards. The share with 3-yr sec. school increased through the whole period.
Stylised model of market for university educated workers: D1 D2 D3 S1 S2 1.Shift in supply dominates. w ,E 2.Shift in demand dominates, w , E S3
Supply of university/college educated increased through the whole period. Agrees with decrease in returns to schooling in the earlier period but not with increase afterwards. There must have been a shift in demand with effects that dominated over those of increased supply. ◦ (to be continued below)
P90/P10P90/P50P50/P Earnings dispersion according to the LLS
Both Sweden and the US saw big increases in dispersion over the 1980s and 1990s. Larger increase in the US and from a higher level. In both countries dispersion increased in both ends of the distribution but more in the upper part. Both differentials between skill groups (returns to schooling and to experience) and within them (residual distribution) increased.
The sharp increase in earnings inequality was not a uniform international development. Big increases also in UK (at least to the mid-1990s), Australia, New Zealand and others. BUT Very little change or even decreases over either the 80s, 90s or both in Germany, France, Japan, Norway and other countries.
Decreasing returns to education shifted to increasing. In Sweden the decrease in gender differentials slowed down or stopped. Increasing relative wages for youth in the 1970s and 1980s, falling in the1990s. (Can be due to selection effects – fewer young people work and more of them have part- time ”extra” jobs.)
That requires a strong increase in demand. Candidates: A. Increased international specialisation. i.Through trade – imports have higher content of low skilled labour and exports lower ii.Through capital movements – production which requires a lot of unskilled labour moves where it is cheapest. (But today highly skilled work, like programming is moved too.) B. Technological change biased towards skilled labour – the IT revolution. But these affected all OECD countries – yet the increase in dispersion wasn’t uniform.
Sweden: 1950s-70s – a strong union movement which tried to limit inequality, particularly among blue-collar workers (the solidaristic wage policy). Highly centralised bargaining and agreements until After that more decentralised and individualised wage bargaining. Agrees in time with increase in dispersion. Research shows connection between egalitarian ambitions of unions and wage compression, particularly for blue-collar workers. There were cut-backs in the public sector and public sector wages declined relative to those in the private sector. US: From the 1980s, both (many) employers tried to restrict unionisation and union influence and so did government policies in the Reagan period. The minimum wage decreased substantially in real terms in the 1980s.
The institutional changes have probably had more impact on the lower part of the wage distribution and demand shifts more on the upper. A lot of research with sometimes conflicting results – and a lot more that remains to be done.
Wages change over the workers life-time. ◦ They increase with experience and according to human capital theory they start from the lowest level and increase fastest in job with educational content. Dispersion of life-time earnings is smaller than dispersion in one particular year. This has been shown with panel data in many countries but some country differences remain – the US wage distribution is particularly unequal both in the long- and short term.
The children of high earners earn more than the children of low earners. High earner parents make sure their children get a good education But given education, there is still an effect. ◦ Biological explanations ◦ Social explanations (cultural and social capital)
The elasticity of the son’s wage with respect to the father’s in US studies is often 0.3 – 0.4 Comparable studies in Sweden and the US found an elasticity of 0.4 in the US, 0.25 in Sweden. (Jäntti & Björklund) Public funding of education and periods of expansion of education tend to lower the intergenerational coefficient (increase social mobility).