Transcript between countries

International inequality
(Concepts 1 and 2)
Milanovic, “Global inequality and
its implications”
Lectures 3-5
Definitions
Three concepts of inequality defined
Concept 1 inequality
Concept 2 inequality
Concept 3 (global) inequalty
Definitions
Different types of inequality
Individuals in:
Countries
World
Countries in:
The usual within-country
distributions
(e.g. inequality in the US is
greater than in Sweden)
-----
Global income distribution:
distribution of persons in the
world
(comparable prices)
Distribution of countries’
GDP per capita
(rich vs. poor countries; are
the poor countries catching up
or not;
the convergence literature)
(comparable prices)
What world inequality are we talking about?
Comparison between the three concepts of inequality
Main source of
data
Unit of
observation
Welfare
concept
National
currency
conversion
Within-country
distribution
(inequality)
Concept 1:
unweighted
inter-national
inequality
National
accounts
Country
GDP or GNP
per capita
Concept 2:
weighted international
inequality
National
accounts
Country
(weighted by its
population)
GDP or GNP
per capita
Concept 3:
“true” world
inequality
Household
surveys
Individual
Mean per capita
disposable
income or
expenditures
Market exchange rate or PPP exchange rate (but
different PPP concepts used)
Ignored
Ignored
Included
Concept 1 inequality,
1950-2002
Year
19
19
19
19
19
19
19
19
19
19
19
19
19
19
98
96
94
92
90
88
86
84
82
80
78
76
74
72
100
70
68
66
64
62
60
58
56
54
52
50
140
19
19
19
19
19
19
19
19
19
19
19
Number of countries or coverage of world population
Inequality between countries
coverage: countries and
population
160
Number of countries included
120
Coverage of world population (in %)
80
60
40
20
0
About 140 countries included; about 6200 country/year GDPs
almost 100 percent of world population and world GDP (in
current dollars)
current countries projected backward (NEW)
 SIMA World Bank data used to get benchmark 1995 $PPP
GDP per capita; then these GDP per capita projected backward
and forward using countries’ real growth rates (78% of data
from WB sources; others mostly from national SYs; some from
PennWorld Tables, UN sources)
Inequality, 1950-2005:
The mother of all inequality disputes
0.7
Global inequality (Concept 3)
Gini coefficient
0.6
Weighted international inequalty
(Concept 2)
Weighted international
inequality without China
0.5
Unweighted international inequality
(Concept 1)
0.4
According to Concept 1, countries' performances have diverged
over the last two decades
Unweighted inter-national inequality, 1950 to 2000
0.560
0.540
0.520
0.480
World
0.460
0.440
0.420
0.400
0.380
0.360
19
50
19
52
19
54
19
56
19
58
19
60
19
62
19
64
19
66
19
68
19
70
19
72
19
74
19
76
19
78
19
80
19
82
19
84
19
86
19
88
19
90
19
92
19
94
19
96
19
98
20
00
Gini coefficient
0.500
Year
And it is not only because Africa is falling behind
0.560
0.540
0.520
0.480
World
0.460
0.440
0.420
0.400
World without
Africa
0.380
0.360
19
50
19
52
19
54
19
56
19
58
19
60
19
62
19
64
19
66
19
68
19
70
19
72
19
74
19
76
19
78
19
80
19
82
19
84
19
86
19
88
19
90
19
92
19
94
19
96
19
98
20
00
Gini coefficient
0.500
Year
Gini and Theil Index
0.300
Gini and Theil Index
Years
Latin America, Caribbean
0.000
Years
19
51
19
53
19
55
19
57
19
59
19
61
19
63
19
65
19
67
19
69
19
71
19
73
19
75
19
77
19
79
19
81
19
83
19
85
19
87
19
89
19
91
19
93
19
95
19
97
19
99
19
51
19
53
19
55
19
57
19
59
19
61
19
63
19
65
19
67
19
69
19
71
19
73
19
75
19
77
19
79
19
81
19
83
19
85
19
87
19
89
19
91
19
93
19
95
19
97
19
99
Gini and Theil Index
Gini and Theil Index
0.300
19
50
19
52
19
54
19
56
19
58
19
60
19
62
19
64
19
66
19
68
19
70
19
72
19
74
19
76
19
78
19
80
19
82
19
84
19
86
19
88
19
90
19
92
19
94
19
96
19
98
20
00
19
50
19
52
19
54
19
56
19
58
19
60
19
62
19
64
19
66
19
68
19
70
19
72
19
74
19
76
19
78
19
80
19
82
19
84
19
86
19
88
19
90
19
92
19
94
19
96
19
98
20
00
Regional convergence and divergence (Gini and Theil index)
0.600
Africa
0.600
Asia
0.500
0.500
0.400
0.400
0.300
0.200
0.200
0.100
0.100
0.000
0.000
Years
Eastern Europe and former USSR
0.600
0.600
0.500
0.500
0.400
0.400
0.300
0.200
0.200
0.100
0.100
0.000
Years
WENAO
0.600
0.500
Note: Theil Index is always
shown by the lower line. The
definition of the Theil entropy
index is
1  yi  yi
   ln
n  
0.300
0.200
i


0.100
0.000
19
50
19
52
19
54
19
56
19
58
19
60
19
62
19
64
19
66
19
68
19
70
19
72
19
74
19
76
19
78
19
80
19
82
19
84
19
86
19
88
19
90
19
92
19
94
19
96
19
98
20
00
Gini and Theil Index
0.400
Years
where yi = income of i-th
country, μ=mean income, and
n=-sample size.
Concept 1 inequality in historical perspective:
Convergence/divergence during different
economic regimes
60
NEOLIBERAL
FIRST GLOBALIZATION
DEGLOBALIZATI
ON
WAR
DEVELOPMENT
AL STATE
50
40
30
Gini
20
10
Theil
0
1820
1870
1890
1900
1913
1929
1938
1952
1960
1978
2000
Relationship between Gini and ln
GDI per capita, 2002
Asia
Latin America
Eastern Europe
West
Total
80
60
40
20
gini
20
40
60
80
Africa
6
8
10
12 6
8
lngdpppp
Graphs by 5 regions
10
12 6
8
10
12
Convergence and
growth theory
Convergence and divergence
• Unconditional or σ convergence (original studies by
Baumol for OECD countries based on Maddison data).
All countries end up with the same steady-state
equilibirum level (NCGT).
• Slower growth of richer countries as MPK falls and they
get closer to technological frontier (technology is freely
available to all)
• Conditional or β convergence (Barro with human capital
only on the RHS instead of K and L). Growth
regressions; based also on endogeneous (“new”) growth
theory; each country ends with its own steady-state
equilibrium
Relationship between δ and β
convergence
• Sigma (or any other inequality measure)
convergence: direct judgment about dispersal of
income levels
• Beta convergence: indirect (regression-based)
regularity; thus not very useful except that it
allows us to retrieve a structural parameter in a
growth model
• β does not imply δ convergence; but δ
convergence implies β convergence (Wodon and
Yitzhaki 2006 Economic Letters).
• Endogeneous growth in response to increasing
returns to scale (no ↓ MPs), monopolistic
competition (no free competition), and no free
diffusion of technology (all key neoclassical
assumptions abandoned), role of policies and
institutions important
• Noted: Lucas paradox: capital flows from rich to
rich countries; mean country incomes diverge
• But β convergence compatible with greater
dispersal of growth rates and incomes
• Often meaningless: if Ethiopia had education
level and institutions of the US, it would grow
faster than the US! (These factors are
concommitant with high income, not
independent of it.)
State steady income
y*  Ae gt [ s /( n  g    
• Depends on A = initial technology but also resource
endowment, climate, institutions etc, g = rate of
technological progress, s = savings (investment) rate, n
= population or labor growth rate, δ = depreciation, α =
share of capital in total output. y* = income per
effective unit of labor.
• In unconditional convergence, all economies the same,
β<0 even if no other RHS variable
• Or economies differ only if one or more of these
parameters differ. Some of the parameters to be
included on the RHS. And find out if β is negative then.
• But do not forget about A!
Endogenous growth: Romer’s key
points
• Technological change propels growth
• Technological change endogenous (responds to
incentives)
• Technological change (knowledge) is a non-rival
but partially excludable good (non-rival: can be
used by many; excludable: you can exclude
some people from using it)
• Excludability (intellectual property rights) is
crucial to cover fixed costs of research and leads
to increasing economies of scale (and hence to
divergence in incomes)
• (Romer: The origins of endogenous growth, JEP, 1994)
Panel approach : heterogeneity of
countries
• Allow for country-fixed effect (contained in A);
large differences in technology (A): variables like
institutions, climate etc. which are in country
fixed effect influence income level (not sufficient
to use K, L)
• Instrument for A; since A is “kitchen-sink”
variable, it can be instrumented by almost any
variable
• If both A and g differ, no convergence
• If parameter heterogeneity (Pasaran & Binder);
no sense to talk about cross-country regressions
which constrain the parameters (even in panels)
The bottom line
• σ convergence among rich countries since WW2
and possibly earlier; at least in terms of wagerates (Williamson), and even during the Interwar years (Milanovic, Restat)
• σ divergence for the world recently, but also
historically, since the Industrial revolution
• σ or unconditional divergence is the same as
increase in Concept 1 inequality (Gini instead of
st. deviation of logs)
• The world of increasing returns to scale PF is a
world of high income and very high inequality
(examples of Sylicon valley, soccer)
Link between HOS trade, growth convergence and
global income distribution
• Factor price equalization theory depends on
openness (not necessarily trade); HOS affects
factor prices
• Under full openness, factor prices will be equal
across countries
• Yet per capita incomes may differ, since per
capita income = Σ wi (factor prices) where
weights (w) depend on factor endowments
which are not equal across countries
• So trade may affect within-national distributions,
and growth rates of poor and rich countries =>
determining global income distribution
The two periods,
1960-1978 and 19782000
Focus first on inequality between countries:
Discontinuity in development trends around
1978-80
• The watershed years (Bairoch)
• Tripling of oil prices
• Increase in real interest rates (from –1% to
+5% in the USA and the world)
• Debt crisis
• China’s responsibility system introduced
• Latin American begins its “lost decade”, E.
Europe/USSR “stagnate”
But also discontinuity in inequality
trends
Within-country inequalities have been rising
during the last two decades (US, UK, China,
India)
Inequalities between countries are rising since
1978
Population weighted inequality between
countries decreasing since 1978 thanks to
growth in China and India (Caveat: acc. to
Maddison Concept 2 inequality is almost stable)
Inequality among people in the world is very
high (Gini between 62 and 66) but its direction of
change is not clear
The outcome:
• Middle income countries declined (Latin
America, EEurope/former USSR)
• China and India pulled ahead
• Africa’s position deteriorated further
• Developed world pulled ahead
• World growth rate decreased by about 1 %
(compared to the 1960-78 period)
Different way to look at world
growth rates
1960-1980
1980-2000
Unweighted (each country
counts the same)
2.9
0.8
Percentage negative
23
33
China
2.7
8.2
India
1.2
3.6
Population-weighted
3.0
3.2
World
2.6
1.6
Annual per capita growth rates
1980-2002
Mean
Median
Percentage
negative
“Old OECD”
1.9
2.0
17
Middle
income
countries
LLDC
1.0
1.8
33
0.1
0.8
43
Growth over 1980-2002 period as function of
initial (1980) income
Distribution of population (in %; year 2000)
according to how country did over 1980-2000
Africa
Asia
WENAO
LLDC
Big time
winners
(>58%)
Winners
13
90
7
26
34
7
93
27
Losers
44
3
0
38
Big time
losers
(>20%)
9
0
0
9
100
100
100
100
Total
The Four Worlds
Define four worlds:
• First World: The West and its offshoots
• Take the poorest country of the First World
(e.g. Portugal)
• Second world (the contenders): all those
less than 1/3 poorer than Portugal.
• Third world: all those 1/3 and 2/3 of the
poorest rich country.
• Fourth world: more than 2/3 below
Portugal.
The border countries and their GDP per
capita levels (in $PPP, 1995 prices)
1960
1978
2000
Greece (13821)
Barbados
(13297)
Malaysia (9887)
Slovak (8595)
Egypt (4630)
Bulgaria (4313)
First to
second
Portugal (3205)
Croatia (3085)
Second to
third
Haiti (2139)
Malaysia (2120)
Portugal (7993)
Puerto Rico
(7662)
Armenia (5294)
Fiji (5156)
Third to
fourth
Nigeria (1080)
Madagascar
(1031)
Guyana (2728)
Cote d’Ivoire
(2649)
Overall upward and downward mobility
1960-78 and 1978-2000
1978-2000
1960-78
Four Worlds in 1960
Four Worlds in 2003
Four worlds in 1960 and 2003
1960
2003
Number of % of
Number of % of
countries population countries population
First
41
26
27
16
Second
22
12
7
2
Third
39
13
29
37
Fourth
25
49
72
46
Poorer than during Carter
Parts of Africa where 2000 GDI per capita
is less than in 1980 (350m people )
US GDI per
capita in the
meantime
increased 50%
Poorer than during J.F. Kennedy
Parts of Africa where 2000 GDI per capita
is less than in 1963 (180m people )
US GDI per
capita in the
meantime
doubled
Why Concept 1 inequality
matters
• Are poor countries catching up as we would
expect from theory?
• Are similar policies producing the same effects
or not? (Rodrik: convergence of policies,
divergence of outcomes). Why?
• Migration issues
• Countries are not only interchangeable
individuals (random assortments of individuals);
they are cultures. Divergence in outcomes is
elimination of some cultures. Perhaps it’s good,
perhaps not.
3500
Transition countries: continued output
divergence despite policy convergence
4
3000
5
st dev. of gdpppp per capita
6
standard deviation of all EBRD indicators
2500
2
2000
3
standard deviation of GDI per capita
1990
1995
2000
Year...
twoway (line EBRD_sd year) (line gdpppp_sd year, yaxis(2)), legend(off) text(6.2 1997 "standard deviation of all
> EBRD indicators") text(3.5 2000 "standard deviation of GDI per capita")
2005
LAC countries: continued output divergence
despite policy convergence
8.00
4
St deviation of the Lora reform indexindicator
7.00
3.9
6.00
3.8
5.00
3.7
4.00
3.6
St. deviation of GDI per capita
3.00
3.5
2.00
3.4
1.00
3.3
0.00
3.2
1985-1988
1988-1991
1992-1994
1995-1997
1998-1999
The key borders today
• First to fourth world: Greece vs.
Macedonia and Albania; Spain vs.
Morocco (25km)
• First to third world: US vs. Mexico;
Germany vs. Poland; Austria vs. Hungary
In 1960, the only key borders were Argentina and Uruguay (first) vs.
Brazil, Paraguay and Bolivia (third world), and Australia (first) vs.
Indonesia (fourth)
Year 2002
Approximate % of
foreign workers in
labor force
Year 1960
Ratio of real GDI per capita
Greece
(Albanians)
7.5
4 to 1
2.2 to 1
Spain
(Moroccans)
12.0
4.5 to 1
2.3 to 1
United States
(Mexicans)
>10.0
4.3 to 1
3.6 to 1
Austria
(former
Yugoslavs)
Malaysia
(Indonesians)
10.0
2.7 to 1
2.6 to 1
>14.0
5.3 to 1
1.5 to 1
The two periods of international growth
Period
Mean
(unweighted)
incomes: “Rest
against West”
Regional homogeneity
1960-1978
Rest catching up Strong divergence in Asia &
Africa; divergence in
EEurope/FSU; mild
convergence in WENAO
and LAC
1978-2000
All falling behind
except Asia
Continued strong
divergence in Africa, joined
by EEurope; mild
divergence in Asia & LAC;
continued convergence in
WENAO only
Concept 2 inequality,
1950-2000
Moving to Concept 2: its relevance
and irrelevance
• Once we have Concepts 1 & 3, Concept 2
is redundant.
• But we have imperfect grasp of Concept 3
inequality => Concept 2 provides a check
on “true” inequality (its lower bound)
• We use it to approximate “true” inequality.
Think, at the limit, of each individual being
a country
40
Population according to income of
country where they live (2000)
India, Nigeria
20
Percent
30
China
10
Brazil, Russia
WEur, Japan
USA
0
Mexico
0
10000
20000
gdp per capita in ppp
histogram gdpppp [w=popu] if year==2000 & gdpppp<32000 & Dcont==1, bin(20) percent ylabel(0(10)40)
30000
Inequality between population-weighted countries
According to Concept 2, there is convergence among countries…
0.780
0.740
Theil
0.700
0.660
0.620
0.580
Gini
0.540
98
96
94
92
90
88
86
84
82
80
78
76
74
72
70
68
66
64
62
60
58
56
54
52
00
20
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
50
0.500
20
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
19
00
98
96
94
92
90
88
86
84
82
80
78
76
74
72
70
68
66
64
62
60
58
56
54
52
50
Gini coefficient
...or maybe there is not
0.600
World
0.560
World without
China
0.520
0.480
World without India and China
0.440
0.400
Alternative Concept 2
calculations
• Alternative growth rates for China (official-World
Bank, Maddison, Penn World Tables)
• Breaking China, India, US, Indonesia and Brazil
into states/provinces (but redistribution within
nations)
• Breaking China into rural and urban parts
(Kanbur-Zhang data set)
• What PPP to use (Geary-Khamis, EKS, Afriat)
Implied China’s GDP per capita in different years
According to different sources
PWT 6.1
Maddison
World Bank
1952
568
627
262
1960
662
785
497
1966
773
879
534
1978
899
1142
754
1988
1703
2119
1676
1999
3319
3803
3867
2000
3642
na
4144
Concept 2 inequality based on different
data and partitions
0.600
WB with regions and
China rural/urban
WB data with
regions
0.550
Maddison
0.500
World Bank
PWT6.1
0.450
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
Concept 2 inequality based on different
data and partitions
0.600
WB with regions and
China rural/urban
WB data with
regions
0.550
Maddison
0.500
World Bank
PWT6.1
0.450
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
How much has Concept 2 inequality
changed (Gini points; 1985-2000)?
Whole
countries
ChIIBus by
states + whole
countries
R/U for China
World Bank
data
-3.3
Maddison
data
-1.9
-4.0
-2.3
-3.3
-1.9
.4
Distribution of people according to income of
country where they live (ln GDPPPP pc;
countries/provinces/states/R-U China, 1980-00
1980
.2
0
.1
kdensity lngdp
.3
2000
6
7
8
9
10
x
kdensity lngdp
kdensity lngdp
twoway (kdensity lngdp [w=popu] if Dgrand2==1 & year==1980 & lngdp<11)
11
.3
.2
.1
0
kdensitylngdp
.4
.5
From one to two poles?
Concept 2 inequality in 1955
and 2000
6
7
8
x
kdensity lngdp
9
10
11
kdensity lngdp
twoway (kdensity lngdp [w=popu] if Dcont==1 & year==1955 & lngdp<11) (kdensity
Concept 2 between
1980 and 2000
Contributes to decline Reverses decline
(equilibrating factors)
(disequilibrating
factors)
• Inclusion of
provinces/states of
• Higher (old) income
China, India, Brazil,
level in China
Indonesia, US (even if
(Maddison) 1.5 points
many within
• Inclusion of
themselves are
rural/urban break up
diverging!) 0.5 point
of China 0.5 points
Result: we shave off half of the Concept 2 decline