Beyond MDG Dashboards: Consideration of Joint

Distribution in Measuring Poverty

Evidence and Measures of Progress in International Development

RSS 2013 International Conference, Newcastle UK

Suman SethSeptember 5, 2013

Outline

• Why is there a need to consider joint distribution and a multidimensional framework for measuring poverty

• The Multidimensional Poverty Index: A Proposal– Methodology– Illustrations

• MPI 2.0 and the post 2015 discussion

What we have: Technical• Increasing data• Improving methodologies

What we need: Policy• Make growth to be inclusive through active

policies • Go beyond income poverty (it is important but

insufficient) • Go beyond dazzlingly complex dashboards of

indicators• Understanding the joint distribution across

deprivations

Path ahead: Ethical and Political• Political critique of current metrics;

exploration • Measures in 2010 HDR sparked interest and

debate• Post-2015 requires re-thinking Data and

Measures

Why New Emphasis on Poverty Measurement?

Economic Growth is Not Always InclusiveIndicators Year India Banglade

sh Nepal

Gross National Income per Capita (in International $)

1990 860 550 510 2011 3620 1940 1260Growth (p.a.) 6.8% 5.9% 4.2%

Under-5 Mortality1990 114.2 138.8 134.62011 61.3 46.0 48.0Change -52.9 -92.8 -86.6

DPT Immunization Rate1990 70 69 432010 72 95 82Change 2 26 39

Adult Pop. with no Education

1990 51.6 55.5 65.82010 32.7 31.9 37.2Change -18.9 -23.6 -28.6

Access to Improved Sanitation (rural pop)

1990 7 34 72010 23 55 27Change 16 21 20

Source: Alkire and Seth (2013). The table is inspired by Drèze and Sen (2011), with minor additions.

Eradicating Income Poverty is not Sufficient (Global Monitoring Report

Progress Status, 2013)

Reduction in income poverty does not reduce other MDG

deprivations automatically. Source: World Bank Data

0

16

32

48

64

80

96

112

128

144

Extreme Poverty Improved Water Primary Completion

Undernourishment Sanitation Infant Mortality

Num

ber o

f Cou

ntries

Target Met Sufficient Progress Insufficient Progress

Moderately Off Target Seriously Off Target Insufficient Data

MDG Dashboards Fail to Reflect Joint Distribution of Deprivations

MDG1

MDG2

MDG3

MDG4

1 0 0 0

0 1 0 0

0 0 1 0

0 0 0 1

An example with four persons (deprived=1, non-deprived=0)

MDG1

MDG2

MDG3

MDG4

0 0 0 0

0 0 0 0

0 0 0 0

1 1 1 1Case 1 Case 2

In both cases, 25% deprived in each MDG indicator

BUT, in Case 2, one person is severely deprived

Motivation for a Multidimensional Approach

• “MDGs did not focus enough on reaching the very poorest” - High-Level Panel on the Post-2015 Development Agenda (2013)– Should be able to distinguish poorest from the less

poor. How?– Deprived in many dimensions simultaneously?

• “Acceleration in one goal often speeds up progress in others; to meet MDGs strategically we need to see them together” - What Will It Take to Achieve the Millennium Development Goals? (2010)– Emphasis on joint distribution and synergies

• “While assessing quality-of-life requires a plurality of indicators, there are strong demands to develop a single summary measure” - Stiglitz Sen Fitoussi Commission Report (2009)– One summary index is more powerful in drawing

policy attention

Value-added of a Multidimensional Approach

What can a meaningful multidimensional measure do?

• Provide an overview of multiple indicators at-a-glance

• Show progress quickly and directly (Monitoring/Evaluation)

• Inform planning and policy design• Target poor people and communities• Reflect people’s own understandings

(Flexible)• High Resolution

– zoom in for details by regions, groups, or dimensions

The Multidimensional

Poverty Index

Alkire Foster Methodology

1. Select dimensions, indicators and weights (Flexible)

2. Set deprivation cutoffs for each indicator (Flexible)

3. Apply to indicators for each person from same survey

4. Set a poverty cutoff to identify who is poor (Flexible)

5. Calculate Adjusted Headcount Ratio (M0) – for ordinal data (such as MDG indicators),

– Reflects incidence, intensity

Sabina Alkire and James Foster, J. of Public Economics 2011

Multidimensional Poverty Index (MPI)

An adaptation of Alkire and Foster (2011) which can deal with the binary or categorical data and was introduced by Alkire and Santos (2010) and UNDP (2010)

A person is identified as poor using a counting approach in two steps1) A person is identified as deprived or not in each dimension using a set of deprivation cutoff2) Based on the deprivation profile, a person is identified as poor or not

Terms: deprived and poor are not synonymous

How is MPI Computed?

The MPI uses the Adjusted Headcount Ratio:

H: The percent of people identified as poor, it shows the incidence of multidimensional poverty

A: The average proportion of deprivations people suffer at the same time; it shows the intensity of people’s poverty

Alkire, Roche, Santos, and Seth (2013)

.

Formula: MPI = H × A

One implementation of the Global MPI (104 countries): Dimensions, Weights &

Indicators

3 Dimensions

10 Indicators

Years of Schooling

(1/ 6)

School Attendance

(1/ 6)

Education (1/ 3)

Child Mortality

(1/ 6)

Nutrition

(1/ 6)

Health (1/ 3) Standard of Living (1/ 3)

Coo

king

Fue

l

Sani

tation

Wat

er

Ele

ctrici

ty

Flo

or

Ass

et O

wne

rshi

p

(1/ 18 Each)

Identify Who is PoorA person is multidimensionally poor

if she is deprived in 1/3 of the weighted indicators.

(censor the deprivations of the non-poor)

33.3%

39%

Properties Useful for Policy

15

The MPI

• Can be broken down into incidence (H) and the intensity (A)

• Is decomposable across population subgroups– Overall poverty is population-share weighted average of

subgroup poverty

• Overall poverty can be broken down by dimensions to understand their contribution

What Kind of Policy

Analysis Can be Done?

Country A:

Country B:

Policy Relevance: Incidence vs. Intensity

50.00

55.00

60.00

65.00

70.00

75.00

50.00

51.00

52.00

53.00

54.00

55.00

56.00

57.00

58.00

59.00

60.00

0.30

0.31

0.32

0.33

0.34

0.35

0.36

0.37

0.38

0.39

0.40

0.41

0.42

Before

MultidimensionalHeadcount

(H)

Intensity of Deprivations

(A)

Multidimensional Poverty Index(MPI = H * A)

50.00

55.00

60.00

65.00

70.00

75.00

50.00

51.00

52.00

53.00

54.00

55.00

56.00

57.00

58.00

59.00

60.00

0.30

0.31

0.32

0.33

0.34

0.35

0.36

0.37

0.38

0.39

0.40

0.41

0.42

Before

MultidimensionalHeadcount

(H)

Intensityof Deprivations

(A)

Multidimensional Poverty Index(MPI = H * A)

50.00

55.00

60.00

65.00

70.00

75.00

50.00

51.00

52.00

53.00

54.00

55.00

56.00

57.00

58.00

59.00

60.00

0.30

0.31

0.32

0.33

0.34

0.35

0.36

0.37

0.38

0.39

0.40

0.41

0.42

After

Before

MultidimensionalHeadcount

(H)

Intensity of Deprivations

(A)

Multidimensional Poverty Index(MPI = H * A)

50.00

55.00

60.00

65.00

70.00

75.00

50.00

51.00

52.00

53.00

54.00

55.00

56.00

57.00

58.00

59.00

60.00

0.30

0.31

0.32

0.33

0.34

0.35

0.36

0.37

0.38

0.39

0.40

0.41

0.42

After

Before

MultidimensionalHeadcount

(H)

Intensity of Deprivations

(A)

Multidimensional Poverty Index(MPI = H * A)

Policy oriented to the poorest of the poorPoverty reduction policy (without inequaliy focus)

Source: Roche (2013)Country B reduced the intensity of

deprivation among the poor more. The final index reflects this.

Policy Relevance: Incidence vs. Intensity

Bangladesh 2004MPI=0.365

Bangladesh 2007MPI=0.289

Nepal 2006MPI=0.350

Nepal 2011MPI=0.217

48%

49%

50%

51%

52%

53%

54%

55%

40% 45% 50% 55% 60% 65% 70% 75% 80%

Inte

nsity

(A)

Incidence (H)

Very similar annual reduction in MPI

Alkire and Roche (2013)

India (1999-2006): Uneven Reduction in MPI across Population Subgroups

19-0.110 -0.090 -0.070 -0.050 -0.030 -0.010

Urban (*) [0.116]

Rural (*) [0.368]

General (*) [0.229]

OBC (*) [0.301]

SC (*) [0.378]

ST (*) [0.458]

Sikh (*) [0.115]

Christian () [0.196]

Hindu (*) [0.306]

Muslim () [0.32]

Absolute Change (99-06) in MPI-I

Stat

es (Si

gnifi

canc

e) [M

PI-

I in

199

9]

Religion

Caste

Slower progress for Scheduled Tribes (ST)

and Muslims

Alkire and Seth (2013)

Reduction in MPI across Indian States

20We combined Bihar and Jharkhand,

Madhya Pradesh and Chhattishgarh, and Uttar Pradesh and Uttarakhand

Stronger reductions

in Southern

states

Slower reductions in initially

poorer states

Comparison with Change in Income Poverty Headcount Ratio

(p.a.)

21

-3.50%-3.00%-2.50%-2.00%-1.50%-1.00%-0.50%0.00%0.50%

Change in MD Poverty (k = 1/3) Change in PCE Poverty

Dimensional Breakdown Nationally?

22

-12.0%

-10.0%

-8.0%

-6.0%

-4.0%

-2.0%

0.0%

Abs

olut

e Cha

nge

in C

H R

atio

Indicator (Statistical Significance) [1999 CH Ratio]

Dimensional Breakdown in Six States?

23

Distribution of Intensities among the Poor

Madagascar (2009)MPI = 0.357

H = 67%

Rwanda (2010)MPI = 0.350

H = 69%

The MPI 2.0 and the Post-2015 discussion

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Nig

erE

thio

pia

Mal

iBu

rund

iBu

rkin

a Fa

soLi

beria

Gui

nea

Som

alia

Moz

ambi

que

Sier

ra L

eone

Sene

gal

DR

Cong

oBe

nin

Uga

nda

Rwan

daTi

mor

-Les

teM

adag

asca

rM

alaw

iTa

nzan

iaZ

ambi

aCh

adM

aurit

ania

Cote

d'Iv

oire

Gam

bia

Bang

lade

shH

aiti

Togo

Nig

eria

Indi

aCa

mer

oon

Yem

enPa

kist

anK

enya

Lao

Cam

bodi

aN

epal

Repu

blic

of C

ongo

Nam

ibia

Zim

babw

eLe

soth

oSa

o To

me

and

Prin

cipe

Hon

dura

sG

hana

Van

uatu

Djib

outi

Nic

arag

uaBh

utan

Gua

tem

ala

Indo

nesi

aBo

livia

Swaz

iland

Tajik

ista

nM

ongo

liaPe

ruIra

qPh

ilipp

ines

Sout

h A

frica

Para

guay

Chin

aM

oroc

coSu

rinam

eG

uyan

aE

ston

iaTu

rkey

Egy

ptTr

inid

ad a

nd T

obag

oBe

lize

Syria

n A

rab

Repu

blic

Colo

mbi

aSr

i Lan

kaA

zerb

aija

nM

aldi

ves

Kyr

gyzs

tan

Dom

inic

an R

epub

licH

unga

ryCr

oatia

Vie

t Nam

Mex

ico

Czec

h Re

publ

icA

rgen

tina

Tuni

sia

Braz

ilJo

rdan

Uzb

ekis

tan

Ecu

ador

Ukr

aine

Mac

edon

iaM

oldo

vaU

rugu

ayTh

aila

ndLa

tvia

Mon

tene

gro

Pale

stin

ian

Terri

torie

sA

lban

iaRu

ssia

n Fe

dera

tion

Serb

iaBo

snia

and

Her

zego

vina

Geo

rgia

Kaz

akhs

tan

Uni

ted

Ara

b E

mira

tes

Arm

enia

Bela

rus

Slov

enia

Slov

akia

Comparing the Headcount Ratios of MPI Poor and $1.25/ day Poor

Intensity 69.4% & More Intensity 50-69.4% Intensity 44.4-50% Intensity 33.3-44.4% $1.25 a day

MPI vs. $1.25-a-day

Height of the bar: MPI Headcount RatioHeight at ‘•’ : $1.25-a-day Headcount Ratio

Measuring the Post-2015 MDGs

What we found from Global MPI - $1.25/poverty and MPI do not move

together- MPI reduction is often faster than

$1.25/day poverty- Political incentives from MPI are more

direct

-5

-4

-3

-2

-1

0

1

2

3

Ann

ualiz

ed abs

olut

e va

riat

ion

MPI Incidenece $1.25 incidence

Measuring the Post-2015 MDGs

28

Create an MPI 2.0 in post 2015 MDGs (Alkire and Sumner 2013)

- To complement $1.25/day poverty- To reflect interconnections between

deprivations- To track ‘key’ goals using data from same

survey- To celebrate success

Note: MPI is not a Composite Index like the HDI or the HPI

Multidimensional Poverty Index - MPI

• Shows joint distribution of deprivations (overlaps)

• Changes over time: informative by region, social group, indicator (inequality)

• National MPIs: tailored to context, priorities

• MPI 2.0: comparable across countries• National MPI and Global MPI 2.0 can be

reported like national income poverty and $1.25/day

• Data needs: feasible – use 39 of 625 questions in DHS

Published: in annual Human Development Report of UNDP

Method: Alkire and Foster 2011 J Public Economics Examples: see www.ophi.org.uk

The Global Multidimensional Poverty Peer Network (Global

MPPN)

Angola, Bhutan, Brazil, Chile, China, Colombia, ECLAC, Ecuador, El Salvador, Dominican Republic, Germany, India, Iraq, Malaysia, Mexico, Morocco, Mozambique, Nigeria, OECD, the Organization of Caribbean States, OPHI, Peru, Philippines, SADC, and Vietnam

Joined by: President Juan Manuel Santos of Colombia

Nobel Laureate Amartya Sen

Launched: June 6, 2013

The Global Multidimensional Poverty Peer Network (Global

MPPN)• On 24 September, 2013: event in the United

Nations N Lawn Conf room 7• Attendees: Ministers from Philippines, Nigeria,

Mexico, Colombia, El Salvador, the Secretary of State of Germany, President of Colombia, Head of DAC at OECD, and others

• Subject: Speak on an MPI 2.0– The Network has decided to advocate a MPI 2.0 as part

of the post-2015 process as a measure of income poverty is not enough, and nor is a dashboard.