2014.10.21 - naec seminar_skills-inequality-well-being

SKILLS DISTRIBUTION, WAGE INEQUALITY AND SOCIAL

INEQUALITY

Dirk Van DammeOECD Directorate for Education and Skills (EDU)

New Approaches to Economic ChallengesSeminar, 21 October 2014

Marco PaccagnellaOECD Directorate for Education and Skills (EDU)

• Overall purpose of this seminar is to provide evidence on how the distribution of skills (and not only the average) relates to various outcome measures:– Wage inequality– Social inequality (Gini coefficient)– Economic output (GDP/capita)

The distribution of human capital matters

• Analysis of OECD Survey of Adult Skills (PIAAC) data (2012)– Focusing on numeracy as a critically

important foundation skill– Focusing on cross-country variation

The distribution of human capital matters

SKILLS AND WAGE INEQUALITY

• Marco Paccagnella

• EDU/SBS

• High levels of inequality are a huge political concern, especially in the midst of a prolonged recession

• Human capital is a crucial factor affecting the growth potential of an economy…

• …but how does it affect inequality?– Increasing returns to education?

– Skill-biased technical change?

– Which role for economic institutions?

Motivation

• Joint analysis of the distribution of skills (numeracy proficiency) and (labour) earnings

• Estimates of the returns to education and proficiency along the distribution of wages

• Decomposition of cross-country differences in wage inequality

This talk

• PIAAC: unique dataset with comparable individual-level information on education, proficiency, and wages

• Drawback: can’t look at the household level

• Preferred measure of dispersion: percentile ratios/differences

The data

• Consistent ranking of countries, irrespective of the indicator and the domain (literacy/numeracy)– High dispersion in US, FR, ES, CA– Low dispersion in JP, KR, SK, CZ

• Bottom-end inequality generally higher than top-end inequality

The distribution of proficiency

Inequality indices – Numeracy SkillsCountry CV 90th-10th 90th-50th 50th-10th

Australia 0.21 136.59 62.33 74.26Austria 0.18 121.24 55.95 65.30Canada 0.21 138.28 62.61 75.67Czech Republic 0.16 110.94 50.90 60.03Denmark 0.18 126.10 57.47 68.63Estonia 0.17 113.92 53.46 60.45Finland 0.18 127.65 59.21 68.44France 0.22 141.80 62.39 79.41Germany 0.20 133.09 59.10 73.99Ireland 0.21 129.33 59.28 70.05Italy 0.20 126.26 59.87 66.39Japan 0.15 110.05 50.89 59.17Korea 0.17 114.60 51.31 63.29Netherlands 0.18 125.11 53.97 71.14Norway 0.19 131.77 57.90 73.88Poland 0.20 127.86 59.20 68.66Slovak Republic 0.17 117.16 51.03 66.12Spain 0.21 129.61 57.08 72.53Sweden 0.20 132.84 58.74 74.10United States 0.23 144.84 66.66 78.18

       Flanders (Belgium) 0.18 127.84 57.13 70.71England/N. Ireland (UK) 0.21 137.71 64.38 73.33         OECD Average 0.20 130.99 59.20 71.79

Dispersion in Numeracy

Inequality indices – Literacy SkillsCountry CV 90th-10th 90th-50th 50th-10th

Australia 0.18 122.28 55.04 67.24Austria 0.16 110.10 50.51 59.60Canada 0.18 125.56 56.20 69.36Czech Republic 0.15 102.34 47.06 55.28Denmark 0.18 116.25 49.88 66.38Estonia 0.16 111.84 50.99 60.85Finland 0.18 123.49 55.16 68.33France 0.19 123.94 54.03 69.91Germany 0.18 121.56 54.40 67.16Ireland 0.18 115.71 52.14 63.57Italy 0.18 113.74 53.68 60.05Japan 0.13 99.78 44.06 55.72Korea 0.15 103.81 46.31 57.49Netherlands 0.17 121.61 51.88 69.73Norway 0.17 115.30 49.98 65.32Poland 0.18 120.92 55.11 65.81Slovak Republic 0.15 99.37 42.89 56.49Spain 0.19 123.52 55.30 68.22Sweden 0.18 122.24 52.86 69.38United States 0.18 126.13 57.14 68.99entities        Flanders (Belgium) 0.17 119.07 51.11 67.96England/N. Ireland (UK) 0.18 123.45 57.03 66.41         OECD Average 0.18 119.41 53.29 66.12

Dispersion in Literacy

Distribution of Numeracy Proficiency0

.00

2.0

04

.00

6.0

08

.01

0 100 200 300 400 500

United States Czech RepublicFrance JapanOECD Average

FRANCEJAPAN

Distribution of (log) Hourly Wages

FRANCE

JAPAN

Skills and Wage Inequality

Top-end Inequality

Bottom-end Inequality

• Focus on groups defined by age and education

• Important to know how homogeneous such groups are in terms of proficiency and earnings

Within-group Dispersion

Skill Inequality Declines with Education

Wage Inequality Does Not

Skill Dispersion Increases with Age

As Does Wage Dispersion

• Unconditional Quantile Regression: Estimate the impact of changing the distribution of explanatory variable on the marginal quantiles of the outcome variable – wages, in our case

• If the estimated impact of a variable is larger at the top than at the bottom of the distribution, then an increase in that variable is associated with an increase in inequality

The Drivers of Wage Inequality

• We focus on the impact of years of education and numeracy

• At the same time, we control for basic socio-demographic characteristics

• This method also allow to decompose cross-country differences in wage inequality into “quantity” and “price” components

The Drivers of Wage Inequality

Basic Results

Cross-country Heterogeneity

Which Kind of Education?

Analysis at the Country Level

Analysis at the Country Level

Analysis at the Country Level

Analysis at the Country Level

Analysis at the Country Level

Analysis at the Country Level

Analysis at the Country Level

Analysis at the Country Level

Analysis at the Country Level

Analysis at the Country Level

Analysis at the Country Level

Analysis at the Country Level

Analysis at the Country Level

Analysis at the Country Level

• We take the United States as a reference country

• We ask how much of the higher level of inequality in the United States can be explained by differences in “quantities” (i.e. in the population distribution of certain characteristics) vs. “prices” (i.e. differences in the way such characteristics are rewarded in the labour market)

A Decomposition Exercise

• It is mostly an accounting exercise, which disregards “general equilibrium effects”

• It involves a number of arbitrary decisions (but results seem to be robust to those)

• As a consequence, should be taken with an appropriate degree of caution

A Decomposition Exercise

Country Raw Gap

90/10

Composition Effect Wage Structure Effect

  Education Numeracy Total Education Numeracy Total

Australia 0.507 -0.022 -0.023 -0.074 0.459 0.060 0.581

Czech R. 0.624 0.035 0.024 0.068 -0.268 0.844 0.556

Denmark 0.701 0.009 -0.009 0.016 0.628 0.377 0.685

France 0.715 0.066 0.000 0.086 0.312 0.232 0.629

Germany 0.218 0.001 -0.018 0.019 0.662 0.191 0.199

Italy 0.441 0.114 -0.013 0.156 0.130 1.076 0.285

Japan 0.247 0.038 -0.030 0.010 -0.235 0.233 0.237

Korea -0.121 0.019 0.001 0.164 0.199 0.519 -0.286

Sweden 0.873 0.036 -0.024 0.020 0.382 0.241 0.852

Results – A Snapshot

• Composition effects seem to play a minor role– Proficiency has the “wrong” sign: most

countries are more proficient than the US

– Education has the “right” sign: most countries are less educated than the US

• Wage structure effects account for 30 to 90% of the observed gap

Results

• Can only speculate about relative importance of institutions vs. market forces

• In any case, policies can play a key role in shaping the evolution on inequality and its social impact

• Positive take-home message: investing in skills could raise earnings without causing increases in inequality

Tentative Conclusions / Interpretation

Country Gap 90th/10th

Composition Effect Wage Structure Effect  Education Numeracy Total Education Numeracy Total

Australia 0.507 -0.022 -0.023 -0.074 0.459 0.060 0.581

Austria 0.536 0.046 -0.030 0.054 0.381 0.029 0.482Canada 0.275 0.007 -0.017 -0.022 0.427 0.065 0.297Czech R. 0.624 0.035 0.024 0.068 -0.268 0.844 0.556Denmark 0.701 0.009 -0.009 0.016 0.628 0.377 0.685

England/UK 0.392 0.028 -0.032 0.001 0.131 -0.162 0.391

Estonia 0.117 0.020 -0.013 0.009 0.543 0.276 0.109Finland 0.713 0.041 -0.011 0.061 0.051 0.386 0.652Belgium 0.684 0.028 0.010 0.111 0.321 0.585 0.572France 0.715 0.066 0.000 0.086 0.312 0.232 0.629Germany 0.218 0.001 -0.018 0.019 0.662 0.191 0.199

Ireland 0.381 -0.098 -0.004 -0.088 -0.058 0.168 0.469Italy 0.441 0.114 -0.013 0.156 0.130 1.076 0.285Japan 0.247 0.038 -0.030 0.010 -0.235 0.233 0.237Korea -0.121 0.019 0.001 0.164 0.199 0.519 -0.286Netherlands 0.475 0.004 -0.034 -0.027 0.602 0.143 0.502

Norway 0.743 -0.008 -0.023 -0.028 0.591 0.240 0.771Poland 0.285 0.019 -0.001 -0.065 0.027 0.435 0.350Slovak Rep. 0.263 0.011 -0.007 -0.004 -0.366 0.402 0.268

Spain 0.366 0.072 -0.002 0.084 0.116 0.620 0.282Sweden 0.873 0.036 -0.024 0.020 0.382 0.241 0.852

Results – 90/10 difference

Country Gap 90th/50th

Composition Effect Wage Structure Effect  Education Numeracy Total Education Numeracy Total

Australia 0.269 -0.010 -0.012 -0.030 0.132 -0.375 0.299Austria 0.308 0.039 -0.006 0.068 -0.051 -0.254 0.239Canada 0.244 -0.005 0.005 -0.010 0.561 -0.036 0.254Czech R. 0.394 0.022 0.006 0.015 -0.389 -0.074 0.378Denmark 0.471 0.017 -0.002 0.018 0.010 -0.190 0.453England/UK 0.195 -0.002 -0.003 -0.002 0.323 -0.223 0.197

Estonia 0.094 -0.018 0.000 -0.024 0.482 -0.163 0.118Finland 0.378 0.022 0.004 0.041 -0.105 -0.130 0.337Belgium 0.390 0.019 0.012 0.073 -0.024 -0.046 0.317France 0.346 0.051 0.000 0.072 -0.076 -0.334 0.274Germany 0.276 0.001 0.003 0.040 0.123 -0.124 0.235Ireland 0.176 -0.017 -0.000 -0.003 0.135 -0.191 0.179Italy 0.240 0.063 -0.005 0.115 -0.080 0.047 0.124Japan 0.065 0.015 0.006 0.039 -0.117 -0.120 0.026Korea -0.080 0.000 0.008 0.116 0.259 0.152 -0.196Netherlands 0.323 0.005 0.023 0.036 0.045 0.059 0.287

Norway 0.428 -0.003 -0.003 -0.003 0.203 -0.194 0.432Poland 0.136 -0.002 0.002 -0.129 0.349 -0.074 0.264Slovak Rep. 0.145 0.006 -0.012 -0.028 -0.320 -0.292 0.173

Spain 0.187 0.011 -0.003 0.037 0.177 0.072 0.149Sweden 0.446 0.023 -0.018 0.007 0.027 -0.340 0.439

Results – 90/50 difference

Country Gap 50th/10th

Composition Effect Wage Structure EffectEducation Numeracy Total Education Numeracy Total

Australia 0.238 -0.011 -0.012 -0.044 0.328 0.435 0.282Austria 0.229 0.007 -0.024 -0.014 0.432 0.284 0.243Canada 0.030 0.012 -0.022 -0.013 -0.134 0.100 0.043Czech R. 0.231 0.012 0.018 0.053 0.120 0.918 0.177Denmark 0.230 -0.008 -0.007 -0.001 0.617 0.567 0.231England/UK 0.196 0.031 -0.029 0.003 - 0.192 0.061 0.194

Estonia 0.023 0.039 -0.013 0.033 0.061 0.439 -0.009Finland 0.335 0.020 -0.015 0.020 0.155 0.516 0.315Belgium 0.293 0.009 -0.001 0.038 0.346 0.631 0.255France 0.369 0.014 0.000 0.014 0.395 0.567 0.355Germany -0.057 -0.000 -0.021 -0.021 0.539 0.315 -0.036Ireland 0.205 -0.081 -0.004 -0.084 -0.193 0.359 0.289Italy 0.201 0.051 -0.009 0.041 0.209 1.029 0.161Japan 0.182 0.023 -0.035 -0.029 -0.119 0.353 0.211Korea -0.041 0.018 -0.007 0.048 -0.059 0.367 -0.089Netherlands 0.151 -0.001 -0.057 -0.063 0.557 0.084 0.214

Norway 0.315 -0.004 -0.020 -0.024 0.388 0.434 0.339Poland 0.149 0.021 -0.003 0.063 0.509 -0.542 0.086Slovak Rep. 0.118 0.005 0.005 0.023 -0.046 0.695 0.095

Spain 0.180 0.061 0.001 0.047 -0.061 0.549 0.133Sweden 0.427 0.013 -0.006 0.014 0.355 0.581 0.413

Results – 50/10 difference

SKILLS, SOCIAL INEQUALITY AND

ECONOMIC OUTPUT

• Dirk Van Damme

• EDU/IMEP

• How is the distribution of numeracy skills in the adult population related to overall social inequality as measured by the Gini coefficient?– And how are they related to economic

output as measured by GDP per capita

• And how are skills distributions in particular groups related to overall social inequality and economic output?

Questions

  GiniGDP per capita

Mean score -.63 .11

Percentage of adults scoring at or below Level 2 .59 -.20

Percentage of adults scoring at Level 4 or 5 -.54 .36

Percentile difference 90th minus 10th percentile .35 .61

Percentile difference 75th minus 25th percentile .40 .59

Percentile difference 90th minus 50th percentile .47 .49

Percentile difference 50th minus 10th percentile .21 .65

Overview of country-level correlations of various numeracy distribution measures with Gini and GDP/capita

A higher mean numeracy score is positively related to higher social equality

245 250 255 260 265 270 275 280 285 2900.20

0.25

0.30

0.35

0.40

Australia

Austria

Canada

Czech RepublicDenmark

Estonia

Finland

GermanyIreland

ItalyJapan

Korea

Netherlands

Norway

Poland

Slovak Republic

Spain

Sweden

United States

Flanders (Belgium)

United KingdomR² = 0.388766537010996

Mean numeracy score

Gini coefficient

But a wide skills dispersion is not very strongly related to higher social inequality…

105 110 115 120 125 130 135 140 1450.20

0.25

0.30

0.35

0.40

Australia

Austria

Canada

Czech Rep Denmark

Estonia

Finland

GermanyIreland

ItalyJapan

Korea

Netherlands

Norway

Poland

Slovak Rep

Spain

Sweden

United States

Flanders

UK

R² = 0.119482758733097

Score point difference between percentile 90 and 10 on the numeracy scale

Gini

…while a wider skills dispersion relates positively with higher economic output

110 115 120 125 130 135 140 14515000

20000

25000

30000

35000

40000

45000

50000

AustraliaAustria Canada

Czech Rep

Denmark

Estonia

Finland

GermanyIreland

Italy

Japan

Korea

Netherlands

Norway

Poland

Slovak Rep

Spain

Sweden

United States

Flanders UK

R² = 0.366289883536397

Score point difference between percentile 90 and 10 on the numeracy scale

GDP per capita

A higher skills dispersion at the top of the distribution relates positively to higher social inequality…

50 52 54 56 58 60 62 64 66 680.20

0.25

0.30

0.35

0.40

Australia

Austria

Canada

Czech Rep Denmark

Estonia

Finland

GermanyIreland

ItalyJapan

Korea

Netherlands

Norway

Poland

Slovak Rep

Spain

Sweden

United States

Flanders

UK

R² = 0.225365468034165

Score point difference between percentile 90 and 50 on the numeracy scale

Gini

…as well as to higher economic output

50 52 54 56 58 60 62 64 66 6815000

20000

25000

30000

35000

40000

45000

50000

AustraliaAustria Canada

Czech Rep

Denmark

Estonia

Finland

GermanyIreland

Italy

Japan

Korea

Netherlands

Norway

Poland

Slovak Rep

Spain

Sweden

United States

Flanders UK

R² = 0.241261161380775

Score point difference between percentile 90 and 50 on the numeracy scale

GDP per capita

But the relationship between higher skills dispersion in the lower half and higher social inequality is much weaker…

58 63 68 73 780.20

0.25

0.30

0.35

0.40

Australia

Austria

Canada

Czech Rep Denmark

Estonia

Finland

GermanyIreland

ItalyJapan

Korea

Netherlands

Norway

Poland

Slovak Rep

Spain

Sweden

United States

Flanders

UK

R² = 0.0490393941452107

Score point difference between percentile 50 and 10 on the numeracy scale

Gini

…while there still is a strong relationship with economic output

58 63 68 73 7815000

20000

25000

30000

35000

40000

45000

50000

AustraliaAustria Canada

Czech Rep

Denmark

Estonia

Finland

GermanyIreland

Italy

Japan

Korea

Netherlands

Norway

Poland

Slovak Rep

Spain

Sweden

United States

Flanders UK

R² = 0.413273337293026

Score point difference between percentile 50 and 10 on the numeracy scale

GDP per capita

More low-skilled adults relates positively to higher social inequality…

35 40 45 50 55 60 65 70 750.20

0.25

0.30

0.35

0.40

Australia

Austria

Canada

Czech RepDenmark

Estonia

Finland

Germany Ireland

ItalyJapan

Korea

Netherlands

Norway

Poland

Slovak Rep

Spain

Sweden

United States

Flanders

UKR² = 0.347930052361252

Percentage adults scoring below Level 2 on the numeracy scale

Gini

…while more high-skilled adults relates negatively to social inequality

4 6 8 10 12 14 16 18 200.20

0.25

0.30

0.35

0.40

Australia

Austria

Canada

Czech Rep Denmark

Estonia

Finland

GermanyIreland

ItalyJapan

Korea

Netherlands

Norway

Poland

Slovak Rep

Spain

Sweden

United States

Flanders

UK

R² = 0.291572239929386

Percentage adults scoring Level 4 or 5 on the numeracy scale

Gini

  GiniGDP per capita

Score-point difference 16-24 year-olds minus 55-64 year-olds

-.01 -.36

Score-point difference between adults with tertiary and lower than upper secondary education

.17 .43

Score-point difference 75th minus 25th percentile among adults with lower than upper secondary education

.19 .38

Score-point difference 75th minus 25th percentile among adults with upper secondary education

.30 .53

Score-point difference 75th minus 25th percentile among adults with tertiary-type A education

.36 .32

Score-point difference between adults with at least one parent who attained tertiary education and adults with neither parent who attained upper secondary education

.19 .25

Country-level correlations of various skills distribution measures among specific groups and Gini and GDP/capita

A higher skills distribution among mid-educated adults is positively related to economic output

50 52 54 56 58 60 62 64 66 6815000

20000

25000

30000

35000

40000

45000

50000

AustraliaAustria Canada

Czech Rep

Denmark

Estonia

Finland

GermanyIreland

Italy

Japan

Korea

Netherlands

Norway

Poland

Slovak Rep

Spain

Sweden

United States

Flanders UK

R² = 0.28104281737691

Score point difference between percentile 75 and 25 on the numeracy scale - upper secondary education attainment

GDP per capita

• A higher mean numeracy level, more high-skilled and less low-skilled are all related to less social inequality

• However, the width of the skills distribution don’t seem to matter a lot for social inequality

• But its shape matters: a wider dispersion in the upper half is related to higher social inequality

• A wider dispersion of skills seems also to be related to higher economic output

Tentative conclusions

• Improving skills of adults seems to be good for economic output, even when such policies widen the total skills distribution

• While this does not seem to harm social equality a lot, except when the better-skilled distance themselves from the median

• But leaving behind a large group of low-skilled adults is also bad for social equality

Tentative conclusions

Thank you !

dirk.vandamme@oecd.orgmarco.paccagnella@oecd.org

www.oecd.org/edu/ceri www.oecd.org/site/piaac/ twitter @VanDammeEDU

64