Inequality in Britain:
an explanation of recent trends
Luke Sibieta,
November 23rd, 2011
© Institute for Fiscal Studies
An economic approach to inequality
The gap between “rich” and “poor”
But how should we measure it?
Why should we care about inequality?
Can we explain recent trends?
How should we measure it?
Inequality of what?
– Household income, earnings, wages, consumption,
material living standards, utility?
Compare different points in the income distribution
Summary measures of the entire distribution with
desirable properties
– E.g. Gini Coefficient, Mean Log Deviation
Substantial rise in inequality during the 1980s
Gini Coefficient
0.40
0.35
0.30
0.25
0.20
Source: IFS
© Institute for Fiscal Studies
Top of income distribution races away
6.00
Ratio
5.00
4.00
3.00
2.00
1.00
90/10
Source: IFS
© Institute for Fiscal Studies
50/10
99/50
Gini – Mid 80s
Gini - Mid 2000s
Source: OECD. Figures not directly comparable with those on other slides. Mid 80s Germany refers to West Germany.
© Institute for Fiscal Studies
OE
CD
ex
ic
o
M
US
A
Ire
la
nd
tra
lia
Au
s
Ita
ly
UK
ad
a
Ca
n
Ja
pa
n
an
ce
Fr
Ge
rm
an
y
0.50
0.45
0.40
0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
Sw
ed
en
Gini Coefficient
The Gini: international comparisons
Real earnings growth
Sweden (1980-2005)
Avg annual
earnings growth
3.0
2.5
2.0
1.5
1.0
0.5
0.0
-0.5
-1.0
Source: OECD
Notes: Full-time male workers only
© Institute for Fiscal Studies
3.0
2.5
2.0
1.5
1.0
0.5
0.0
-0.5
-1.0
USA (1980-2005)
France (2000-2005)
Avg annual
earnings growth
Avg annual
earnings growth
3.0
2.5
2.0
1.5
1.0
0.5
0.0
-0.5
-1.0
Avg annual
earnings growth
UK (1980-2005)
3.0
2.5
2.0
1.5
1.0
0.5
0.0
-0.5
-1.0
Why should we care about inequality?
Economics largely focuses on ‘Efficiency’ and
‘Pareto optimality’, says very little about ‘Equity’
Efficiency considerations for valuing inequality
– Social unrest and crime, trust and social capital,
long-term generational consequences
– But some benefits as a result of incentive effects
Welfare considerations of a social planner
– What is the optimal welfare metric?
Why has inequality gone up?
There are many competing explanations for
changes in inequality, including:
–
–
–
–
Changing industrial structure, etc.
Returns to education
Regional changes (North/South gap)
Demographic Change
Statistical decompositions allow us to weigh up
some of these competing explanations
Results taken from Brewer, Muriel and WrenLewis (2010)
Inequality – subgroup decomposition
Inequality measures tell us how
unequally incomes are distributed
Lowest
Income
Highest
Inequality – subgroup decomposition
But suppose we divide the population
into subgroups (e.g. ethnic groups, or
employed vs. unemployed)
Lowest
Income
Highest
Inequality – subgroup decomposition
We see that some groups have very high
inequality (e.g. the self-employed)
Lowest
Income
Highest
Inequality – subgroup decomposition
And others have low
inequality, with low
average incomes
(etc.)
Lowest
While others have low
inequality, but high average
incomes (e.g. the employed)
Income
Highest
Inequality – subgroup decomposition
Increasing inequality could be caused by:
Inequality increasing
within some groups:
Lowest
= “Within-group inequality”
effect
Income
Highest
Inequality – subgroup decomposition
Or increasing inequality between groups
Lowest
Income
Highest
Inequality – subgroup decomposition
Or increasing inequality between groups
(i) because their average
incomes change
Lowest
Income
= “Relative income
effect”
Highest
Inequality – subgroup decomposition
Or increasing inequality between groups
(ii) because more people join
a group with high inequality
Lowest
Income
= “Population change
(within)” effect
Highest
Inequality – subgroup decomposition
Or increasing inequality between groups
(iii) because more people
join groups with low (or
high) incomes
Lowest
= “Population change
(mean)” effect
Income
Highest
Subgroup decomposition example: by
employment status of head of household
We divide the population into groups according to
the schooling of the head of household:
• Left education aged 16 or under
• Left education aged 17-19
• Left education aged 20+
• Unknown or still in education
Decompose inequality changes into the four effects
described earlier
Subgroup decomposition – by education level
Between-group inequality
Period
Change in
overall
inequality
(I0)
Withingroup
inequality
1978 - 84
18
13
1984 – 88
42
34
1988 – 91
18
16
1991 – 95-96
–12
–13
1995-96 – 2000-01
11
10
2000-01 – 04-05
–9
–10
1
1
–2
2004-05 – 06-07
10
10
1
1
–2
Population
change
(within)
Population
change
(mean)
Relative
income
0
1
4
0
2
Mostly
group
0 within
1
Relative
income effect
inequality
important
during
1
1
1980s
1
2
5
1
–2
–2
Subgroup decompositions can be unclear
Suppose we have the following change:
Lowest
Income
Highest
Subgroup decompositions can be unclear
Gender subgroup decomposition....
... would partly attribute it to
decreasing female income
Lowest
Income
Highest
Subgroup decompositions can be unclear
Colour subgroup decomposition....
... would partly attribute to
decreasing yellow income
Lowest
Income
Highest
Subgroup decompositions can be unclear
Which factor is more important?
Lowest
Income
Highest
Subgroup decompositions can be unclear
Which factor is more important?
Factor decomposition allows us to
consider all factors simultaneously
Lowest
Income
Highest
Inequality (log variance x 1000)
Income inequality – factor decomposition
350
Employment
Status
Health
300
Ethnic group
250
Education
200
Age
150
Gender
100
Household
type
Region
50
Residual
0
1968 1972 1976 1980 1984 1988 1992 1996 2000 2004
Source: HBAI Data (FES and FRS) and authors’ analysis
What we do not know (and why it matters)
There is much we do not understand! But the answer matters.
Different explanations for why inequality has increased have
different implications for various ideas of ‘fairness’ or ‘justice’
Studies have examined whether increases in income
inequality reflect :
Higher volatility in people’s incomes from year to year.
Or increases in ‘permanent inequality’ between people
who are always poor and always rich (i.e. reductions in
social mobility).
© Institute for Fiscal Studies
How much can government affect inequality?
Tax and state benefit system clearly makes a difference
But can be very costly to just rely on fiscal redistribution.
Countries with low inequality (e.g. Scandinavia) tend to
have low ‘pre tax and benefit’ inequality; much of recent
rise in inequality due to rise in pre-tax earnings inequality.
Income distribution shaped by the distribution of private
incomes earned in the labour market.
© Institute for Fiscal Studies
How much can government affect labour market
outcomes?
Depends what drives it!
But ultimately, people can not earn a lot in the labour
market unless they have skills that employers value.
It is not surprising that the most equal societies are
different to the most unequal ones from the bottom up,
e.g. early educational achievements, health, etc.
It seems very difficult to only target equality in specific
things (e.g. ‘income’), because everything is connected!
© Institute for Fiscal Studies