Underlying themes. 2. Income inequality over the ... Underlying themes: ⢠Study of inequality should be central ... US
Opening Conference of the GINI Project
March 2010
Income inequality in historical and comparative perspective
A B Atkinson, Nuffield College, Oxford and LSE
1
1. Underlying themes 2. Income inequality over the twentieth century in OECD countries 3. Seeking explanations
2
Underlying themes: • Study of inequality should be central to social science; • Both long-run and comparative perspectives are necessary: - Rich source of evidence - Living in “unusual times” - Countries not all moving in step
• We need to be more serious about data: - Not rely on “stylised facts” - Consider full range of possible data, varying across time and across countries - Data quality not (0,1) 3
Here focus on a narrow concept of inequality: money income at a point in time. Income includes both earnings AND property income and transfers. MISSING (among other elements) • Social dimensions of inequality; • Public goods; • Life-course and inequality of opportunity.
4
1. Underlying themes 2. Income inequality over the twentieth century in OECD countries 3. Seeking explanations
5
US and UK: the end of a special relationship? Figure A Income inequality in UK and US 1908-2008 51
US BEA synthetic estimate GROSS US CPS GROSS US CPS new GROSS UK Blue Book synthetic estimate NET UK Economic Trends NET US top 1% UK top 1% UK top 0.05%
Gini
Gini
20
Share of top 1% 36
10
Gini 31
Share of top 0.05%
Gini
26
5
Share of top 1%
Share of 15 top 1%
41
0
19 01 19 05 19 09 19 13 19 17 19 21 19 25 19 29 19 33 19 37 19 41 19 45 19 49 19 53 19 57 19 61 19 65 19 69 19 73 19 77 19 81 19 85 19 89 19 93 19 97 20 01 20 05
Gini coefficient %
46
25
6
Patchwork of data, but can draw some conclusions: the US and UK are not simple cases of U-Turn • Similar pre-WW2 but differ afterwards • US has long plateau from 1945 to 1970s • US U more like • UK had falling inequality from 1945 to late 1970s • In the UK inequality rose much more in the 1980s than in the US • In the UK, overall inequality levelled off since 1990, but top shares continued to rise. • overall inequality in UK more like
7
Canada/US more similar, but still differences Figure F Income inequality in Canada and US 1913-2008
of top 1%
20 US BEA synthetic estimate GROSS US CPS GROSS US CPS new GROSS Statistics Canada US top 1% Canada top 1%
Gini
16 Share of top 1% 12
49
44
Gini
Gini
39
8
4
Step change? 34 19 01 19 05 19 09 19 13 19 17 19 21 19 25 19 29 19 33 19 37 19 41 19 45 19 49 19 53 19 57 19 61 19 65 19 69 19 73 19 77 19 81 19 85 19 89 19 93 19 97 20 01 20 05
Gini coefficient %
54
Share of top 1%
Share 59
0
8
Common experience in the Nordics? Figure C Three Nordic countries in Detail 34.0 SWE 1967
Gini
31.5 SWE 1
Gini coefficient %
29.0
SWE 2 SWE 3
26.5
24.0
FI NOR
21.5
19.0 1901 1906 1911 1916 1921 1926 1931 1936 1941 1946 1951 1956 1961 1966 1971 1976 1981 1986 1991 1996 2001 2006
9
Nordic recent trends: SWEDEN “Alongside the higher economic standard of households since the mid-1990s, differences in income have also increased. The difference has not been as large since Statistics Sweden began these measurements in 1975” (Statistics Sweden press release May 2009). NORWAY “All income classes enjoyed a strong rise in household income [in 2007]. However, people at the top of the income distribution had the strongest rise” (Statistics Norway press release March 2009). FINLAND “Growth of income differentials continued in 2007” (Statistics Finland, May 2009). 10
IMPORTANCE of CAPITAL INCOMES
Two neighbours Figure D Germany and Netherlands 1901-2007 50
NL 1 gross
35 Note: changing geographical boundaries of Germany
30
Gini
25
35
20
30
15
25
10
DE top 1%
20
5
NL top 1%
DE survey
DE SOEP panel study
19 01 19 06 19 11 19 16 19 21 19 26 19 31 19 36 19 41 19 46 19 51 19 56 19 61 19 66 19 71 19 76 19 81 19 86 19 91 19 96 20 01 20 06
Gini coefficient %
40
Share of top 1%
Share of top 1%
45
NL 2 tax units disposable income NL 3 households disposable income DE DIW synthetic estimate
Germany Gini ↑ 4-5 percentage points 1999-2007
11
Summary: 10 OECD countries (not all shown) • General fall up to 1945 in top income shares and in overall inequality (where evidence available); • Fall in top shares and overall inequality continued up to 1970s in some countries (UK, Nordics, Netherlands) but was less marked in US and in France top shares not greatly changed while overall inequality fell; • Rise in top shares post-1980 in US, Canada and UK; rise less obviously continuing in Nordics; rise less steep or non-existent in Continental Europe; • In nearly all countries Gini is higher today than in 1980 but the extent of the rise varies: if we take the US as a benchmark (+ 5 percentage points), then Canada, Norway and UK are similar, but the increase is larger in Finland and Sweden, followed (recently) by Germany, and is smaller (less than 2 percentage points) or non-existent in France, Italy and the Netherlands; • It is not evident whether rise in Gini is continuing or was a step change: U or U? • Differing coverage of capital incomes. 12
1. Three themes 2. Income inequality over the twentieth century in OECD countries 3. Seeking explanations
13
Different research strategies EVIDENCE: • Examine variation in summary inequality measures (such as Gini) for total income across countries and over time (Panel of Countries approach); • Examine variation in inequality for total income over time, and compare findings for different countries (not pooled); • Examine variation in shares of different income groups (such as top 1%, top 5%, bottom 20%) over time and across countries; • Summarise country distributions by (several) parameter functional forms, and seek to explain differences/changes in these parameters; • Estimate individual income (earnings) equations from micro-data and then examine their implications for aggregate inequality. HYPOTHESES: • Data-driven; • Model-driven.
14
Data-suggested hypotheses: (1) Kuznets Figure A* Income inequality in UK and US
Figure B* Three Nordic countries
51
25
Gini
34.0
Share of top 1% 36
10
Gini
Share of top 0.05%
5
Gini
25
29.0
20
26.5
15
24.0
10
Share of top 1% 21.5
26
0
5
19.0
0
19 0 19 1 0 19 5 0 19 9 1 19 3 1 19 7 21 19 2 19 5 2 19 9 3 19 3 37 19 4 19 1 4 19 5 4 19 9 5 19 3 5 19 7 61 19 6 19 5 6 19 9 7 19 3 77 19 8 19 1 8 19 5 8 19 9 9 19 3 9 20 7 0 20 1 05
1901190619111916192119261931193619411946195119561961196619711976198119861991199620012006
Figure E* France
Figure D* Germany and Netherlands 50
40
30
30
45
25
Gini
25
35
20
30
15
25
10
Gini coefficient %
Gini coefficient %
50
Note: changing geographical boundaries of Germany
Share of top 1%
45
35
Gini Share of top 1% NL NL
40
20
35
15
Share of top 1% 30
10
Share of top 1% Prussia/DE 5
25
15
19 01 19 06 19 11 19 16 19 21 19 26 19 31 19 36 19 41 19 46 19 51 19 56 19 61 19 66 19 71 19 76 19 81 19 86 19 91 19 96 20 01 20 06
19 01 19 06 19 11 19 16 19 21 19 26 19 31 19 36 19 41 19 46 19 51 19 56 19 61 19 66 19 71 19 76 19 81 19 86 19 91 19 96 20 01 20 06
20
5
Share of top 1%
31
31.5
Share of top 1%
Share of 15 top 1%
41
Gini coefficient %
20
Share of top 1%
Gini coefficient %
Gini
Gini
46
30
Data-suggested hypotheses: (2) Financial crises
Share of top 1 per cent in Singapore 1947 to 2007 20
Percentage share of total household income
18
Rise following Asian financial crisis
Commodities boom
16
Independence
14 12 10
Decline from 1950s to early1960s
8 6
Broad stability from mid 1960s to 1990s
4 2 0 1947
1952
1957
1962
1967
1972
1977
1982
1987
1992
1997
2002
2007
Source: Table 5.1.
16
Data-suggested hypotheses: (3) Recessions: Gini coefficients
17
Modelling Components of disposable income Literature has largely focused on individual earnings distribution.
Earnings (8) + Capital income (1) + Transfers - Taxes
Models of wealth accumulation and transmission = minority interest.
Political economy models of welfare (2) state and progressive taxation. 18
Distribution of property income by decile groups
1. FACTOR SHARES
Distribution of earned income by decile groups (of whole population)
BRINGING THE COMPONENTS TOGETHER 2. RANK CORRELATION (copula )
3. INTRODUCE TAXES AND TRANSFERS 19
Conclusions •
The distribution of income is complex phenomenon, even before we look at wider dimensions of inequality! • No simple letter (U or V) summary of recent developments in individual OECD countries; • Differing experience across countries, and between income groups within countries; • Historical evidence provides perspective, and suggests hypotheses, but cautions against extrapolation; • Can approach modelling in different ways; • Useful to consider individual components, and not to forget property income (although measurement problems); • But essential to bring components together.
20