1
A B Atkinson, University of Oxford
Global Economy Lecture 2009
The global distribution of income: past trends and future prospects
2
“Controversy centres on whether inequality has increased or decreased in the recent past. The direction and magnitude of change have been highly charged questions with some authors arguing that globalization has benefited the rich disproportionately, while others argue that it has reduced world income inequalities” (Anand and Segal, page 57).
S Anand and P Segal, “What do we know about global income inequality?”, Journal of Economic Literature 2008
Reasons for disagreement:• Different concerns• Data shortcomings• Different movements at different points in the distribution
3
1. Conceptual issues: what are we concerned with?
2. The international distribution of income: convergence following divergence?
3. Within-country inequality over the twentieth century: a Great U-turn?
4. The global distribution: combining between- and within-country inequality
5. Underlying explanations
6. Where are we headed?
4
1. Conceptual issues: what are we concerned with?
2. The international distribution of income: convergence following divergence?
3. Within-country inequality over the twentieth century: a Great U-turn?
4. The global distribution: combining between- and within- country inequality
5. Underlying explanations
6. Where are we headed?
5
Distribution among whom?International
Distribution (among countries)
•Country A
•Country B
•Country C …
World
Distribution (among people)
WORLD
CITIZENS
Household incomes
GDP per head
6
Distribution of what?
STANDARD ANSWER = Potential living standards, measured by annual disposable income among households adjusted for household composition and for differences in purchasing power across countries and across time.
Inequality typically measured by standard inequality measures, such as Gini coefficient
Implies that international distribution is an intermediate step, not a final concern
7
Distribution of what?
ALTERNATIVE ANSWER = Economic power of countries measured by total national income measured at exchange rates (what can buy on world market)
Measured by concentration measures (such as share of top 12 countries)
Implies that international distribution is a final concern
8
1. Conceptual issues: what are we concerned with?
2. The international distribution of income: convergence following divergence?
3. Within-country inequality over the twentieth century: a Great U-turn?
4. The global distribution: combining between- and within- country inequality
5. Underlying explanations
6. Where are we headed?
DATA: How do we measure the international distribution of income?
Angus Maddison data set on gross domestic product per capita AD1-AD2006
BUT• Official National accounts only started post
1945• Need to compare across time (price indices)• Need to compare across countries (purchasing
power adjustments)
10
Between-country differences: GDP per capita (PPP adjusted) in India and UK
Divergence
Convergence
11
Figure 2 International Lorenz curves 1820-2000
0
10
20
30
40
50
60
70
80
90
100
0 10 20 30 40 50 60 70 80 90 100
Shar
e of
tota
l inc
ome
per c
ent
1820
1900
1950
1980
2000
Diagonal
World population in increasing order of GDP per head of country
12
Top 12 countries concentration ratio
0
20
40
60
80
100
1970 1980 1990 2000 2006
Perc
enta
ge o
f tot
al w
orld
inco
me
in U
S $
TOP 12
exc USSR
Geo-political concerns
13
INTERNATIONAL DISTRIBUTION: Summary
• Long period of divergence following Industrial Revolution – up to mid-century.
• Large countries now becoming middle income.
• BUT leaving poorest countries behind.
• Rich countries as a whole still pulling away.
• Geo-political concentration among countries not falling appreciably.
14
1. Conceptual issues: what are we concerned with?
2. The international distribution of income: convergence followingdivergence?
3.Within-country inequality over the twentieth century: a Great U-turn?
4. The global distribution: combining between- and within- country inequality
5. Underlying explanations
6. Where are we headed?
DATA: How measure the within-country distribution of income?
Household surveys and/or administrative recordsBUT• Surveys relatively recent innovation, and
suffer from non-response;• Administrative data reflect administrative
purpose;• Sources differ across time and across
countries.
16
Look at two aspects of the distribution:
• Overall inequality, measured by Gini coefficient:
A divided by (A+B) Lorenz curve
Gini of 25 per cent means that if take 2 people at random, then on average difference between their incomes is 50 per cent of the average income.
• Share of top 1 per cent in total income.
A B
17Financial crises
Data are patchwork
18Is US U and UK U ?
19Germany Gini ↑ 4-5 percentage points 1999-2007
20
WITHIN-COUNTRY INCOME INEQUALITY (OECD COUNTRIES)
• Long period of substantial decline in inequality, both overall Gini and top income shares.
• Rise in last 30 years in both overall inequality and top shares in Anglo-Saxon countries.
• Later rise in both overall inequality and top shares in Nordic countries and in overall inequality in Germany.
• Upward movement less evident in Netherlands (and France and Italy).
CONCLUDE: Evidence for U, but not to same extent, nor at same time, and need to distinguish U from U.
21
1. Conceptual issues: what are we concerned with?
2. The international distribution of income: convergence followingdivergence?
3. Within-country inequality over the twentieth century: a Great U-turn?
4.The global distribution: combining between- and within- country inequality
5. Underlying explanations
6. Where are we headed?
22
Approached issue by considering separately international distribution and within-country distribution. Need to combine, but not straightforward.
- Need comparable distributional data across countries (no fixed effect)
- Need to go from GDP per capita to average household disposable income. Differences include:
• income paid abroad (Ireland)
• government sector and imputed benefits from government spending (India)
• retained profits of corporations
• income of pension funds
• imputed rent of owner-occupied housing
23
Conclusions on separate elements
∩ +
√
?Probable
24
1. Conceptual issues: what are we concerned with?
2. The international distribution of income: convergence followingdivergence?
3. Within-country inequality over the twentieth century: a Great U-turn?
4. The global distribution: combining between- and within- country inequality
5.Underlying explanations6. Where are we headed?
25
Explanations?
Why no Kuznets inverse-U for within-country inequality?Model of structural change: workers paid more in advanced sector
Second part to Kuznets = concentration of capital and factors counteracting.
∩∩
Early industrialisers
Later industrialisers
26
SINGAPORE Earnings distribution and Top incomes 1965-2006
0
1
2
3
4
5
6
7
8
9
10
1965
1967
1969
1971
1973
1975
1977
1979
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
2005
Rat
io o
f ear
ning
s
0
2
4
6
8
10
12
14
16
18
20
% s
hare
in to
tal g
ross
inco
me
Decile ratio of earnings
Quintile ratio of earnings
Share of top 1% of incomes (RH axis)
Moving outside the OECD
Financial crisis
27
Earnings and Capital shares in UK 1923-2003
0
10
20
30
40
50
60
70
1923 1928 1933 1938 1943 1948 1953 1958 1963 1968 1973 1978 1983 1988 1993 1998 2003
Perc
enta
ge s
hare
of t
otal
wea
lth
0
1
2
3
4
5
6
7
Perc
enta
ge s
hare
of t
otal
ear
ning
s
Long fall in top wealth share
Steep rise in top pay share
Some recovery of top wealth share
Earlier fall in top pay share
Negative shocks to capital but also progressive income and capital taxation.
Earnings and capital income: UK Case Study
28
1. Conceptual issues: what are we concerned with?
2. The international distribution of income: convergence followingdivergence?
3. Within-country inequality over the twentieth century: a Great U-turn?
4. The global distribution: combining between- and within- country inequality
5. Underlying explanations
6.Where are we headed?
29
What can we learn from the past? Between-country distribution
Source: Maddison data for 42 countries
OS
30
5
10
15
20
1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939
FRA
UK
US
CA
AUS
GER
NL
JA
FI
NOR
Within-country distribution 1929-39: Share of top 1% (0.1% in UK) Not all depressed.
31
Gini coefficient and Banking-crisis induced recessions
32
Share of top 1% and Banking-crisis induced recessions
33
Conclusions• We are increasingly thinking in global terms, but we need to clarify underlying source of concern with inequality.• Empirical data greatly improved, but still important gaps in knowledge.• International distribution diverged in a major way, but limited convergence to date.• Within-country inequality fell substantially, but now U-turn.• Grounds for concern about the distributional impact of economic crisis.