Arab countries’ inequality

i a global
in l b l context

Branko Milanovic
Branko Milanovic
ERF Cairo, April 2010
Email: bmilanovic@worldbank org
Email: bmilanovic@worldbank.org

Based on the book Worlds Apart, 2005 and updates
BM note: this
hi iis a f
fully
ll revised
i d lleon2.ppt
2 excludes
l d the
h stuff
ff
on global crisis
 The objectives of the talk
• Review empirical movements in international and 
global inequality 1950‐2010
• Focus on global interpersonal inequality in the last 
twenty years
• Impact of new PPPs globally and on Arab countries
• Show how many household surveys from Arab 
countries are included in WYD database
• Show where in global income distribution are people 
f
from various Arab countries and income groups
i b i di
• Examine within‐national inequality in Arab countries, 
and in a hypothetical community of all Arab nations
di h th ti l it f ll A b ti
1. The background: Global 
inequalities today 
 Three concepts of inequality defined
Concept 1 inequality

Concept 2 inequality

Concept 3 (global) inequality
 Inequality 1950‐2007
The mother of all inequality disputes
The mother of all inequality disputes
China
.75

moves
Concept 3
i
in

Concept 2
.65
5
.55

Divergence ends

C
Concept 1 Divergence
begins
.45

1950 1960 1970 1980 1990 2000 2010

year

With new PPPs
Graph in interyd\dofiles\defines.do
 Concept 1 and Concept 2 inequality
Concept 2
65
.6

India as new
engine of
equalization
..6

Concept 2 without China

..55

Concept 1
.5
.45

1940 1960 1980 2000 2020

year

Graph in interyd\dofiles\defines.do
 All three concepts using Theil (1) coefficient

1.1
Concept 3
1 .9
Theil(1) coeffiicient
.8

Concept 2
.6 .7

Concept 1
.5

1950 1960 1970 1980 1990 2000 2010

year

twoway (line global_theil2 year if contcod=="USA") (line global_theil1 year if contcod=="USA") /*/ (scatter global_theil3 year if
contcod=="USA", msize(vlarge)), ylabel (0.5(0.1) 1.1) xlabel(1950(10)2010) text(0.80 1970 "Concept 2") text(0.52 1975 "Concept 1") text(1.05
1995 "Concept 3") ytitle (Theil(1) coefficient) legend(off)
Use defines.do with gdppppreg.dta
 All three concepts using market exchange rates

.85
Concept 3
.8 .75
nt

Concept 2
Gini coefficien
.65
5 .7

Concept 1
.6
.55

1970 1980 1990 2000 2010

year

Using two_concepts_exrate.do and global_new2.dta

 The difficulty of intuition re. evolution of 
p q y
Concept 3 inequality stems from 
contradictory movements
(1) Greater inequality within nations
(1) G t i lit ithi ti
(2) Greater differences between countries’ 
mean incomes (unconditional divergence 
between 1980 and 2000)
(3) But catching up of large and poor countries 
(China and India)
All of these forces determine what happens to 
GLOBAL INEQUALITY (but they affect it
GLOBAL INEQUALITY (but they affect it 
differently)
 The impact of new PPPs
• Concept 2 inequality increased by almost 10 
Gini points (a level shift)
Gini points (a level shift)
• Somewhat steeper decline of Concept 2 
inequality in the last decade (because India
inequality in the last decade (because India 
and China now appear poorer)
• About 5 Gini points increase in Concept 1 
About 5 Gini points increase in Concept 1
inequality (shift effect; no trend  effect)
• About 5 Gini points increase in global 
About 5 Gini points increase in global
inequality (Concept 2 increases more than 
Concept 3
Concept 3            smaller overlap
smaller overlap as mean 
as mean
incomes “move” further apart)
 (cont.)
• World
World poorer than we thought, Asia in 
poorer than we thought Asia in
particular
• Inequality (in all formulations) greater
Inequality (in all formulations) greater
• Growth rates not affected in WDI but will be 
affected in PWT; so “the past will now 
ff d i PWT “h ill
change” (like in Orwell)
• Two engines of “global equalization”: China 
and India 
 Pattern of change in estimated price levels:  increases in poorer 
and more populous countries (both highly sign. in a regression)
p p ( g y g g )

3
Zaire
2.5

Ethiopia
e level

China
cchange in price
2

India
1.5

Brazil
IDN
1
.5

6 7 8 9 10 11
lngdppp before the change

twoway (scatter price_level_change lngdpppp_old [w=totpop], yline(1)) (qfit price_level_change lngdpppp_old [w=totpop]), text(2.3 8.5
"China") text(1.7 7 "India") text(1.4 8.8 "Brazil") text(1.3 8.2 "IDN") text(3 6.5 "Zaire") text(2.35 6.5 "Ethiopia") ytitle(change in price level)
xtitle(lngdppp before the change) legend(off)
From graph2.do
 Ratio of the new country price levels to the old

Africa Asia Latin America

ZAR
1 1.5 2 2.5

SYR
ETH
MOZ KHM NIC
price level

GIN CPV CHN

PHL
BGD
ZAF HTI
GNB
BFA IND LKA
NER UGAMAR
MRT
COM
SLEMWI SEN NPL
VNM JOR PRYBRA
CMR COL
MLI IDN THA HND DOM
CRIARG
CHL
URY-U
MDGTCD EGY PAK MYS JPN
HKG PER
SLV
GTM MEX
ZMB
NGA
CIV
BEN LAO IRN KOR JAM
evel over old p

SGP BOL PAN

ECUVEN
TZA GAB
.5

COG

6 8 10 12

Eastern Europe West

new price le
1 1.5 2 2.5
5

BIH
ARM
UZBLVA
KGZ
UKR SVK
EST IRLLUX
MKD
GEOLTU
ROMHRV SVN ISR
AUS
TUR GRCNOR
DNK
SWEUSA
TJK
MDA BGR
RUS
KAZPOL
HUN
OU
CZE ITA
CAN
ESP
DEU
CHE
FIN
FRA
AUT
NLD
GBR
BEL
ALBBLR
SRB
.5

6 8 10 12 6 8 10 12
lngpppp before changes
Graphs by 5 regions

twoway scatter price_level_change lngdpppp_old, ylabel(0.5(0.5)2.5) yline(1) mlabel(contcod) /*

*/ by(region) ytitle(new price level over old price level) xtitle(lngpppp before changes)
From graph2.do
 How much have GDP per capita have changed as
result
lt off new price
i level
l l estimates?
ti t ?

300
change in estimated GDP per capita
200

YEM
100

LBN
SAU

EGY
SDN SYR
0

JOR
MAR TUN
COM
-100
0

6 7 8 9 10 11
GDP per capita using old ppp

twoway (scatter change lnwdi1_gdpppp if Darab==0, yline(1) msymbol (oh) msize(large)) (scatter change lnwdi1_gdpppp if
Darab==1, msize(vlarge) mlabel(contcod) legend(off) xtitle(GDP per capita using old ppp) ytitle(change in estimated GDP per
capita))
From icp_2005.dta
2. More details
 Population coverage in WYD dataset
Population coverage in WYD dataset
1988 1993 1998 2002 2005
Africa 48 76 67 77 78
Asia 93 95 94 96 94
E.Europe 99 95 100 97 91
LAC 87 92 93 96 96
WENAO 92 95 97 99 99
World 87 92 92 94 93
Non‐triviality of the omitted countries (Maddison vs. WDI)
 GDI (US dollar) coverage in WYD
GDI (US dollar) coverage in WYD
1988 1993 1998 2002 2005
Africa 49 85 71 71 68
Asia 94 93 96 95 90
E. Europe
E. Europe 99 96 100 99 93
LAC 90 93 95 95 98
WENAO 99 96 96 100 100
World 96 95 96 98 96
 Number of surveys (C‐based)
Number of surveys (C based) in WYD
in WYD
1988 1993 1998 2002 2005
Africa 14(11) 30(27) 24(24) 29(29) 31(29)

Asia 19(10) 26(18) 28(20) 26(18) 22(15)

EEurope 27(0)
( ) 22(0)
( ) 27(14)
( ) 25(16)
( ) 26(25)
( )

LAC 19(1) 20(4) 22(2) 21(1) 18(0)

WENAO 23(0) 23(0) 21(3) 21(2) 22(0)

World 102(22) 121(52) 122(63) 122(66) 119(69)

 Global inequality (with 2005 PPPs)
(distribution of persons by $PPP or US$ income per capita) 

1988 1993 1998 2002 2005

International dollars
Gini i
Gi 68.3
68 3 69.9
69 9 69.4
69 4 70.6
70 6 69.9
69 9
index (2.0) (1.4) (1.8) (1.3) (1.6)
Between 
B t 61.6 62.3 61.7 63.0 61.6
component

US dollars
US dollars
Gini  77.8 80.4 79.6 81.0 79.8
i d
index (1.5) (1.4) (1.3) (1.1) (1.1)
 More than fifty‐fifty world (2005; new PPPs)
Cumulative % of world 
C l ti % f ld Cumulative % of PPP world 
C l ti % f PPP ld IIn a single 
i l
population income/consumption country 
(Germany 05)
5 0.14 1.3
10 0.44 3.3
25 1.9 11.1
50 6.6 28.9
80 25.0 60.1
90 45 75
Top 10 55 25
p
Top 5 36.5 18.4
Top 1 13.4 5.8
 How big is a Gini of 70? (Year 2005, 2005PPPs)

T
Top B
Bottom R i
Ratio
In PPP dollars

5 percent  38% 0.24% 165‐1

10 percent 58% 0.6% 95‐1

In current  $
$

5 percent 45% 0.15% 300‐1

10 percent 67.5% 0.45% 150‐1

10 t
10 top countries
ti 39 115
39,115 570 68 1
68‐1
3. Arab countries in the global context
 What Arab countries are included
i WYD d
in dataset? ?
Egypt All five years

Jordan, Morocco Four years (no 2005 data)

Tunisia, Mauritania, Comoros Three years (no 2002, 2005 data

for TUN, MRT)

Algeria, Yemen Two years

United Arab Emirates, Djibouti, One year only

Syria

Bahrain, Iraq,
Bahrain Iraq Kuwait
Kuwait, Lebanon
Lebanon, No data at all
Libya, Oman, Saudi Arabia,
Somalia, Sudan
Very weak representation of Arab countries
(th lowest
(the l t off allll regions)
i )
• Arab population included in WYD data in 1993,
1998, 2002 ranged between 120 and 140 million,
while total population of the Arab league
countries is about 340 million (in 2007).
• The inclusion is only around one-third (lower
than Sub-Saharan Africa)
• All surveys are expenditure- or consumption-
b
based d ((no iincome surveys). )
• In 2005, the inclusion is even worse, but I have
nott finished
fi i h d th the collection
ll ti off ddata.
t
 Inequality in Arab countries vs. the rest of the
world
ld (2002
(2002-2005)
2005)
60
50

COM
Gini
40

MAR
MRT JOR
SYR

EGY
30
20

6 8 10 12
GDI per capita in 2005 PPP dollars

twoway (scatter gini lngdpppp if Darab==1 & gini<60, mlabel(contcod) msize(vlarge) xscale(log)) (scatter gini lngdpppp if Darab==0 & gini<60,
msize(vlarge) msymbol (oh) xscale(log) legend(off) ytitle(Gini) xtitle(GDI per capita in 2005 PPP dollars
¾)) || qfitci gini lngdpppp
Using world2002_2005.dta
 Where are globally different income groups
off Arab
A b countries’
t i ’ populations?
l ti ?
100
1

Germany
Spain
bution
80
d income distrib

M
Morocco

Jordan
Egypt
60
perrcentile of world
20 40

1 5 10 15 20
country ventile

Use michele-_graph.do
• Germany and Spain: first-order dominance
over all three Arab countries.
• About ½ of Jordanians better off than the
poorest in Spain, but only ¼ of Egyptians
and Moroccans better off than the poorest
p
Spaniards
• Jordan: first-order
first order dominance over Egypt
and Morocco
• Rich Moroccans richer than the richest
Egyptians (even if the difference is small)
but the poorest Moroccans much poorer
than the poorest Egyptians
 Global income positions and $PPP
incomes compared
Country Poorest Middle ventile Richest ventile Spread: top
ventile to bottom
Brazil 1 (100) 66 (1,900) 98 (26,000) 97
Morocco 16 (360) 56 (1,140) 85 (7,200) 69
E
Egypt
t 31 (530) 58 (1
(1,230)
230) 84 (5
(5,800)
800) 53
Jordan 35 (570) 65 (1,800) 89 (10,000) 64
Syria
y 37 ((610)) 64 ((1,700)) 88 ((9,300)) 51
USA 68 (2,000) 93 (13,600) 100 (70,000) 32
France 75 (3,820) 91 (12,100) 100 (45,900) 25
Germany 78 (3,500) 91 (11,600) 100 (38,000) 22

First number gives global percentile position; second number gives expenditures or
disposable income per household member in 2005 international dollars (based on
2005ICP results). Year 2002 except France 2005. Countries ranked by the position
of the poorest ventile.
 How heterogeneous are Arab countries: the
between- and within-inequality
between within inequality (Gini points)
Inequality Total Share of between
inequality in total
between inequality
(%)
units
USA (50 states) 8 40 ~20
China (29 provinces) 24 40+ ~60
60

Russia (89 subjects of 20 40+ ~50

federation)
EU-15 countries (pre- 10.2 33.4 ~30
enlargement)
EU 27 (post enlargement)
EU-27 23 1
23.1 40 3
40.3 ~57
57

8 Arab countries (1998-2000) 22.1 47.1 ~45

EU data calculated from world2002.dta US from the same source; Russia from R. Yemtsov (Russian_regional.xls);
Arab from my pppinc88_93_98.dta.
• Total inter
inter-personal
personal inequality in 8 Arab countries
(Emirates, Comoros, Egypt, Jordan, Morocco,
Mauritania, Tunisia and Yemen) is 47 Gini
points, some 20% greater than inter-personal
inequality in the EU and USA.
• Less than ½ of that total inequality is due to
differences in mean country incomes. Inequality
between units the same in EU and Arab
countries (but EU has 27 members).
• If we were to include 12 missing Arab countries
countries,
both total and between inequality would increase
substantially.
y
 Capture ratio: relationship between HS means and
per capita
it consumptionti ffrom National
N ti lAAccounts
t
Years 1988-1998 Years 2002-2005

5
SYR
3

4
2
caa pture ra tio

c a p tu re ra tio
3
COM MRT
MRT
YEM
MAR JOR

2
JOR
JOR DZA
MAR COM
MAR
1

DZA
TUN
EGY TUN
TUN
MRT
MAR

1
EGY
EGY
JOR
0

6 7 8 9 10
GDI per capita in 2005 PPP dollars 5 6 7 8 9 10
GDI per capita in 2005 PPP dollars

twoway (scatter scale2 lngdpppp if Darab==1 & scale2<4, mlabel(contcod) msize(vlarge) xscale(log)) (scatter scale2 lngdpppp if Darab==0 &
Dexpend==1, msize(vlarge) msymbol (oh) yline(1, lpattern(dash)) xscale(log) legend(off) ytitle(capture ratio) xtitle(GDI per capita in 2005 PPP dollars))
Using pppinc88_93_98.dta and world2002_2005.dta
• Syria a definite outlier: HS data much
higher than per capita consumption or
even GDP per capita from National
Accounts (may be due to the ER)
• Yemen
Y as wellll (i
(in 1993)
• Egypt tends to have a rather low HS
consumption: only about 50% of the value
from National Accounts
• Other countries close to NA data, or
slightly
g y above ((which is normal))
• Book
Book “Worlds
Worlds Apart: Measuring International 
Apart: Measuring International
and Global Inequality”, Princeton UP, 2005; 
and many additions since
and many additions since

• Email: bmilanovic@worldbank.org
E il b il i @ ldb k

• Website: 
http://econ.worldbank.org/projects/inequality