Globalization and Inequality: Comments on Ravallion

Globalization and Inequality:Comments on Ravallion

Lant Pritchett,

KSG, CGD, WB

13 May 2004

Two points

• Current “globalization” talk is, and is bound to be, nonsense.

• Most economists do not, and should not, believe in PWPL (Poverty with low poverty lines) as a social objective—and this is a defensible value judgment—an array of poverty lines is appropriate

Globalization and Inequality

• Real globalization—which included mobility of people--would reduce inequality in well being across spatial regions: full stop.

• We don’t have anything like globalization:– Nation-states have proliferated (more

sovereigns, more borders, more monies)– Movement of people has been off the agenda.

Number of new countries added per year(Source: Braun, Hausmann and Pritchett, “The Proliferation

of Sovereigns”)

0

2

4

6

8

10

12

14

16

18

20

19

43

19

44

19

45

19

46

19

47

19

48

19

49

19

50

19

51

19

52

19

53

19

54

19

55

19

56

19

57

19

58

19

59

19

60

19

61

19

62

19

63

19

64

19

65

19

66

19

67

19

68

19

69

19

70

19

71

19

72

19

73

19

74

19

75

19

76

19

77

19

78

19

79

19

80

19

81

19

82

19

83

19

84

19

85

19

86

19

87

19

88

19

89

19

90

19

91

19

92

19

93

19

94

ssa mena

eca lac

eap other

0.0%

10.0%

20.0%

30.0%

40.0%

50.0%

60.0%

70.0%

1800 1820 1840 1860 1880 1900 1920 1940 1960 1980 2000

Years

Fra

ctio

n o

f T

ota

l In

equ

alit

y D

ue

to

Acr

oss

Co

un

trie

s

Theil coefficient

Mean ln deviation

Fraction of inequality due to across versus within countries

(Source: Bourguignon and Morrison)

2.33.6

4.1 4.4

2.4

6.17.2 7.6

9.0 9.1

0.01.02.03.04.05.06.07.08.09.0

10.0

USA/Ireland 1870

USA/Norway 1870

USA/Sweden 1870

USA/Italy 1870

Spain/Morocco 1990s

USA/Guatemala 1990s

UK/Kenya

Italy/Ethiopia 1990s

NLD/Indonesia 1990s

Japan/Vietnam 1990s

Rat

io o

f w

ages

in

PP

P

With negative region specific shocks of “desired” populations one either gets large changes in growth of populations (ghosts) or growth of

income per head (zombies)

Zombies

Ghosts

Population

Wages

Large movements of population within (large) countries, large differences in income per head

across countries

Why is this graph so facetious?

0

20

40

60

80

100

120

Gai

ns

as %

of

wo

rld

GD

P

All remaining goodsliberalization

3% increase in hostcountry labor force

Full Liberalization oflabor markets

X and inequality

“Does the set of processes called X in which

(a) the number of nation-states rises, but (b) transport costs (including of information) and (c) some nation-states pursue modestly more

liberal policies of some types towards the movement of goods and (perhaps) capital but not labor lead to

Increase or decrease inequality across nations/individuals?

“Globalization” and Y

• It is not clear “globalization” is the key phenomena of the 20th century—nationalism which is deeply anti-global is.

• It cannot be the case that we have an answer as to the impact of “globalization” on Y (where Y=growth, inequality, the environment)—an ill-posed question.

PWLPL vs. SWF

• Does X affect “poverty”?

• FGT(α,PL) requires a poverty line

• Gains to income/expenditure above the PL count zero towards reduction in poverty.

-1.5

-1

-0.5

0

0 0.2 0.4 0.6 0.8 1

Percentile

Headcount

Poverty Gap

Squared Poverty Gap

(to minus infinity)

Watts

EDEI

Contribution of income gains at various points in the income distribution to welfare measures

(Source: Kraay 2004)

Value judgments and poverty

• I personally do not have an objective function in which income/consumption gains count for zero above 1$/day or 2$/day.

• The question is: where can the poverty line be drawn such that zero is even a reasonable approximation?

Arguments against low poverty line

• Non-paternalism—there is no evidence of a sharp reduction in increases in subjective well-being w.r.t. expenditures around 1 or 2 $/day.

• Non-money metric indicators—relationship between income/expenditures and mortality/health/education does not get lower anywhere near 1 or 2 $/day

• Non-money metric indicators indicate the “rich” (upper 20 percent) in poor countries are enormously worse off than the “poor” in rich countries.

Arguments against low poverty line

• Non-paternalism—there is no evidence of a sharp reduction in increases in subjective well-being w.r.t. expenditures around 1 or 2 $/day.

• Non-money metric indicators—relationship between income/expenditures and mortality/health/education does not get lower anywhere near 1 or 2 $/day

• Non-money metric indicators indicate the “rich” (upper 20 percent) in poor countries are enormously worse off than the “poor” in rich countries.

World Values Survey Data on income and “happiness”

0

10

20

30

40

50

60

70

80

90

100

0 5000 10000 15000 20000 25000 30000

GDP per capita, P$1990

Ind

ex o

f su

bje

ctiv

e w

ell-

bei

ng

Arguments against low poverty line

• Non-paternalism—there is no evidence of a sharp reduction in increases in subjective well-being w.r.t. expenditures around 1 or 2 $/day.

• Non-money metric indicators—relationship between income/expenditures and mortality/health/education does not get lower anywhere near 1 or 2 $/day

• Non-money metric indicators indicate the “rich” (upper 20 percent) in poor countries are enormously worse off than the “poor” in rich countries.

Infant mortality has (roughly) the same elasticity w.r.t. GDP per capita over a large range of (average) incomes

1

2

3

4

5

6

4 5 6 7 8 9 10 11

GDP per capita 1990s average (log scale, 1995US$)

Under-

5 m

ort

alty

rate

2000 (

log s

cale

)

Draw the poverty line where MDGs are achieved

Pre

d P

rob

co

mp

letin

g g

rad

e 5

(a

g

Nepal 1995Relative ranking (pc hh expendit

0 .9985181

0

.1

.2

.3

.4

.5

.6

.7

.8

.9

1

Arguments against low poverty line

• Non-paternalism—there is no evidence of a sharp reduction in increases in subjective well-being w.r.t. expenditures around 1 or 2 $/day.

• Non-money metric indicators—relationship between income/expenditures and mortality/health/education does not get lower anywhere near 1 or 2 $/day

• Non-money metric indicators indicate the “rich” (upper 20 percent) in poor countries are enormously worse off than the “poor” in rich countries.

Infant mortality of the richest 20 percent in selected countries

0

20

40

60

80

100

120

140

Brazil

,1996

Philipp

ines(

Cebu)

,91

Vietnam

,92-

93

Nepal,9

6

Nicara

gua,

93

South

Africa

,93

CIV,8

7-88

Ghana

,87-

88

Pakist

an,91

IMR

per

'000

Malnutrition of the richest 20 percent in selected countries

0

5

10

15

20

25

30

35

Brazil

,96

Nicara

gua,

97/9

8

Ghana

,93

CIV,9

4

India

,92/

93

Nepal,9

6

Niger

ia,90

Pakist

an,90

/91

Pro

po

rtio

n s

tun

ted

Arguments against low poverty lines (con’t)

• Revealed preference within and across countries– Poverty lines rise with income both within countries

and over time• There is absolutely no compelling factual or

value argument for low poverty lines.– LPL are not “biologically” based, e(p,Upov)

• Use of LPL is fine, but let us not confuse type I and type II error—almost certainly no one below a LPL is “not poor” but many (billions) are above LPL and are poor by reasonable definitions.

A proposal

• For 1$/day poverty we use the term “destitution”• For 2$/day we use the term “extreme poverty”• Each country uses its own ‘national” poverty line• We use the capital weighted average of the

World Bank shareholder poverty lines—something like 15$/day—to define “global poverty”

World Distribution of Income and “Poverty”

Destitution

Extreme Poverty

National poverty lines

Global Poverty

Poverty is a social construct

• This is not about having a “poverty focus” or not having a “poverty focus” this is about choosing a poverty line.

• For global discussions there is no reason to insist on a low poverty line as the only or even the major metric for success.

• I come not to bury poverty but to raise it.