Poverty Measurement Design Sabina Alkire, Jose Manuel Roche and - - PowerPoint PPT Presentation

Value Judgements in Multidimensional Poverty Measurement Design

Sabina Alkire, Jose Manuel Roche and Maria Emma Santos OPHI Workshop, Oxford, 28 June 2012

Two parts:

• Measurement methodology
• Value Judgements

– Constraints – Complementary Analyses – Uncertainty/Incompleteness – Authority

Measurement methodology

• Why focus on a particular method?

– Focused discussion – Actual examples – Practical Benefits – Clarity

And note:

– Relevance to other methods during field-building

Multidimensional Poverty- our challenge:

• A government would like to create an official

multidimensional poverty indicator

• Desiderata

– It must understandable and easy to describe – It must conform to “common sense” notions of poverty – It must be able to target the poor, track changes, and guide policy. – It must be technically solid – It must be operationally viable – It must be easily replicable

• What would you advise?

Methodology

• Identification – Dual cutoffs
• Aggregation – Adjusted FGT
• All value judgements are assumed given for now:

– Purpose, Dimensions, Indicators, Cutoffs, Weights/Values etc

• Today we focus on one measure M0, which can be

used with ordinal, categorical, binary, or cardinal data. It has been used extensively, for practical reasons.

Review: Unidimensional Poverty

Variable – income Identification – poverty line Aggregation – Foster-Greer-Thorbecke ’84 Example Incomes = (7,3,4,8) poverty line z = 5

Deprivation vector g0 = (0,1,1,0)

Headcount ratio P0 = m(g0) = 2/4 Normalized gap vector g1 = (0, 2/5, 1/5, 0) Poverty gap = P1 = m(g1) = 3/20 Squared gap vector g2 = (0, 4/25, 1/25, 0) FGT Measure = P2 = m(g2) = 5/100

Multidimensional Data

Matrix of well-being scores for n persons in d domains Domains Persons ฀ y  13.1 14 4 1 15.2 7 5 12.5 10 1 20 11 3 1              

Multidimensional Data

Matrix of well-being scores for n persons in d domains Domains Persons z ( 13 12 3 1) Cutoffs ฀ y  13.1 14 4 1 15.2 7 5 12.5 10 1 20 11 3 1              

Deprivation Matrix

Replace entries: 1 if deprived, 0 if not deprived Domains Persons ฀ y  13.1 14 4 1 15.2 7 5 12.5 10 1 20 11 3 1              

Deprivation Matrix

Replace entries: 1 if deprived, 0 if not deprived Domains Persons ฀ g0  1 1 1 1 1 1 1              

Normalized Gap Matrix

Normalized gap = (zj - yji)/zj if deprived, 0 if not deprived Domains Persons z ( 13 12 3 1) Cutoffs These entries fall below cutoffs ฀ y  13.1 14 4 1 15.2 7 5 12.5 10 1 20 11 3 1              

Normalized Gap Matrix

Normalized gap = (zj - yji)/zj if deprived, 0 if not deprived Domains Persons ฀ g1  0.42 1 0.04 0.17 0.67 1 0.08              

Squared Gap Matrix

Squared gap = [(zj - yji)/zj]2 if deprived, 0 if not deprived Domains Persons ฀ g1  0.42 1 0.04 0.17 0.67 1 0.08              

Squared Gap Matrix

Squared gap = [(zj - yji)/zj]2 if deprived, 0 if not deprived Domains Persons ฀ g2  0.176 1 0.002 0.029 0.449 1 0.006              

Identification

Domains Persons Matrix of deprivations ฀ g0  1 1 1 1 1 1 1              

Identification – Counting Deprivations

Domains c Persons ฀ g0  1 1 1 1 1 1 1               2 4 1

Identification – Counting Deprivations

Q/ Who is poor? Domains c Persons ฀ g0  1 1 1 1 1 1 1               2 4 1

Identification – Dual Cutoff Approach

Q/ Who is poor? A/ Fix cutoff k, identify as poor if ci > k Domains c Persons ฀ g0  1 1 1 1 1 1 1               2 4 1

Identification – Dual Cutoff Approach

Q/ Who is poor? A/ Fix cutoff k, identify as poor if ci > k (Ex: k = 2) Domains c Persons ฀ g0  1 1 1 1 1 1 1               2 4 1

Identification – Dual Cutoff Approach

Q/ Who is poor? A/ Fix cutoff k, identify as poor if ci > k (Ex: k = 2) Domains c Persons Note Includes both union (k = 1) and intersection (k = d) ฀ g0  1 1 1 1 1 1 1               2 4 1

Identification – The problem empirically

k = H Union 1 91.2% 2 75.5% 3 54.4% 4 33.3% 5 16.5% 6 6.3% 7 1.5% 8 0.2% 9 0.0%

• Inters. 10

0.0%

Poverty in India for 10 dimensions: 91% of population would be targeted using union, 0% using intersection Need something in the middle.

(Alkire and Seth 2009)

Aggregation

Censor data of nonpoor Domains c Persons ฀ g0  1 1 1 1 1 1 1               2 4 1

Aggregation

Censor data of nonpoor Domains c(k) Persons Similarly for g1(k), etc ฀ g0(k)  1 1 1 1 1 1               2 4

Aggregation – Headcount Ratio

Domains c(k) Persons ฀ g0(k)  1 1 1 1 1 1               2 4

Aggregation – Headcount Ratio

Domains c(k) Persons Two poor persons out of four: H = 1/2 ฀ g0(k)  1 1 1 1 1 1               2 4

Critique

Suppose the number of deprivations rises for person 2 Domains c(k) Persons Two poor persons out of four: H = 1/2 ฀ g0(k)  1 1 1 1 1 1               2 4

Critique

Suppose the number of deprivations rises for person 2 Domains c(k) Persons Two poor persons out of four: H = 1/2 4 3 1 1 1 1 1 1 1 ) (              k g

Critique

Suppose the number of deprivations rises for person 2 Domains c(k) Persons Two poor persons out of four: H = 1/2 No change! Violates ‘dimensional monotonicity’ 4 3 1 1 1 1 1 1 1 ) (              k g

Aggregation

Return to the original matrix Domains c(k) Persons 4 3 1 1 1 1 1 1 1 ) (              k g

Aggregation

Return to the original matrix Domains c(k) Persons ฀ g0(k)  1 1 1 1 1 1               2 4

Aggregation

Need to augment information

deprivation shares among poor

Domains c(k) c(k)/d Persons ฀ g0(k)  1 1 1 1 1 1               2 4 2 / 4 4 / 4

Aggregation

Need to augment information

deprivation shares among poor

Domains c(k) c(k)/d Persons A = average deprivation share among poor = 3/4 ฀ g0(k)  1 1 1 1 1 1               2 4 2 / 4 4 / 4

Aggregation – Adjusted Headcount Ratio

Adjusted Headcount Ratio = M0 = HA Domains c(k) c(k)/d Persons A = average deprivation share among poor = 3/4 ฀ g0(k)  1 1 1 1 1 1               2 4 2 / 4 4 / 4

Aggregation – Adjusted Headcount Ratio

Adjusted Headcount Ratio = M0 = HA = m(g0(k)) Domains c(k) c(k)/d Persons A = average deprivation share among poor = 3/4 ฀ g0(k)  1 1 1 1 1 1               2 4 2 / 4 4 / 4

Aggregation – Adjusted Headcount Ratio

Adjusted Headcount Ratio = M0 = HA = m(g0(k)) = 6/16 = .375 Domains c(k) c(k)/d Persons A = average deprivation share among poor = 3/4 ฀ g0(k)  1 1 1 1 1 1               2 4 2 / 4 4 / 4

Aggregation: Adjusted Poverty Gap

Adjusted Poverty Gap = M1 = M0G = HAG Domains Persons Average gap across all deprived dimensions of the poor: G/ ฀ g1(k)  0.42 1 0.04 0.17 0.67 1              

Aggregation: Adjusted Poverty Gap

Adjusted Poverty Gap = M1 = M0G = HAG = m(g1(k)) Domains Persons Obviously, if in a deprived dimension, a poor person becomes even more deprived, then M1 will rise. Satisfies monotonicity ฀ g1(k)  0.42 1 0.04 0.17 0.67 1              

Aggregation: Adjusted FGT

Consider the matrix of squared gaps Domains Persons ฀ g2(k)  0.422 12 0.042 0.172 0.672 12            

Aggregation: Adjusted FGT

Adjusted FGT is M = m(g(k)) Domains Persons Satisfies transfer axiom ฀ g2(k)  0.422 12 0.042 0.172 0.672 12            

Aggregation: Adjusted FGT Family

Adjusted FGT is Ma = m(ga(t)) for a > 0 Domains Persons Theorem 1 For any given weighting vector and cutoffs, the methodology Mka =(ρk,Ma) satisfies: decomposability, replication invariance, symmetry, poverty and deprivation focus, weak and dimensional monotonicity, nontriviality, normalisation, and weak rearrangement for a>0; monotonicity for a>0; and weak transfer for a>1. ฀ ga (k)  0.42a 1a 0.04a 0.17a 0.67a 1a              

• International MPI
• National Poverty
• Mexico, Colombia
• Well-being
• Bhutan’s GNH
• Adaptations
• Empowerment
• Energy
• Governance

Some Applications

Examples

UNDP’s 2010 Human Development Report first published the MPI

(updated annually for countries with new data)

The MPI (UNDP 2010)

• The MPI 2011 is an international index of acute

multidimensional poverty developed at OPHI

• In 2011 it covers 109 developing countries.
• It was launched in 2010 in the Human Development

Report, and updated in 2011, 2012

• It complements the $1.25/day poverty by bringing

more dimensions into view

• It is the first measure to reflect joint distribution of

disadvantages.

Data: Surveys

Demographic & Health Surveys (DHS - 54) Multiple Indicator Cluster Surveys (MICS - 32)

World Health Survey (WHS – 17)

Additionally we used 6 special surveys covering urban Argentina (ENNyS), Brazil (PNDS), Mexico (ENSANUT), Morocco (ENNVM), Occupied Palestinian Territory (PAPFAM), and South Africa (NIDS) Constraints: Data are 2000-2010, and not all have all 10 indicators.

MPI Dimensions Weights & Indicators

Measurement: Indicators & Cutoffs

• Health

– Child Mortality: If any child has died in the family – Malnutrition: If any interviewed adult in the family has low Body Mass Index; if any child is more than 2 standard deviations below the reference normal weight for age, WHO standards) [WHS has male & female data but no child data; MICS has child data but no adult data; DHS has women 15-49 & child]

Measurement: Indicators & Cutoffs

• Education

– Years of Schooling: if no person in the household has completed 5 years of schooling – Child Enrolment: if any school-aged child is out

• f school, where school-aged is an eight year

period from the national starting age.

Measurement: Indicators & Cutoffs

• Standard of Living

– Electricity (no electricity is deprived) – Drinking water (MDG definitions) – Sanitation (MDG definitions + not being shared) – Flooring (dirt/sand/dung are deprived) – Cooking Fuel (wood/charcoal/dung are deprived) – Assets (deprived if do not own a car/truck and do not own more than one of these: radio, tv, telephone, bike, motorbike, or refrigerator)

Identification: Who is poor?

A person is multidimensionally poor if they are deprived in 33% of the dimensions at the same time.

33%

How do you calculate the MPI?

• The MPI uses the Alkire Foster method:
• H is the percent of people who are identified as poor,

it shows the incidence of multidimensional poverty.

• A is the average proportion of weighted deprivations

people suffer at the same time. It shows the intensity

• f people’s poverty – the joint distribution of their

deprivations.

The MPI is appropriate for ordinal data, and satisfies properties like subgroup consistency, dimensional monotonicity, poverty & deprivation focus. MPI is like the poverty gap measure – but looks at breadth instead – what batters a person at the same time.

Formula: MPI = M0 = H × A

What is new?

Intensity.

The MPI starts with each person, and constructs a deprivation profile for each person. Some people are identified as poor based on their joint

• deprivations. The others are identified as non-poor.
• Most multidimensional poverty measures look at

deprivations one by one, not at the household level.

• Counting measures do look at coupled deprivations but only

provide a headcount, giving no incentive to target those who are deprived in most things at the same time or to reduce intensity.

53 53

54 54

55 55

56

57 57

58 58

59 59

60 60

61

Phuba

Deprived in 67% of dimensions. It doesn’t tell the full story But it gives some idea.

MPI – Key Results

Global Results:

These results are for 109 developing countries, selected because they have DHS, MICS or WHS data since 2000. Special surveys were used for Argentina, Brazil, Mexico, Morocco, Occupied Palestinian Territory, and South Africa They cover 5.3 billion people - 78.6% of the world’s population Of these 5.3 billion people, 31% of people are poor. That is 1.65 billion people.

(2008 population figures taken from Population Prospects 2011; 2010 Revision).

Half of the world’s MPI poor people live in South Asia, and 29% in Sub-Saharan Africa

MPI poor people by region

Total Population in 109 MPI countries

The MPI Headcount Ratios and the $1.25/day Poverty

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

103 of our 109 Countries have income; only 71 have income poverty data within 3 years of MPI. Income data ranges from 1992-2008; MPI from 2000-2010.

Intensity is highest in the poorest countries.

But there is variety…H in High-income countries 1-7%

H in High- and Upper Middle-income countries 1-40%

H in Middle- and High-income Countries 1-77%

H in Low-income Countries ranges from 5-92%

Ghana, Nigeria, and Ethiopia

Ethiopia’s Regional Disparities

Ethiopia

Ethiopia’s Regional Disparities

Addis Ababa Somali Afar Harari Dire Dawa

Nigeria’s Regional Disparities

Nigeria

Nigeria’s Regional Disparities

South West North East

Nigeria

Ghana’s Regional Disparities

Ghana

Ghana’s Regional Disparities

Greater Accra Northern Ghana

CHANGES OVER TIME

Ghana, Nigeria, and Ethiopia

Let us Take a Step Back in Time

Ghana 2003 Nigeria 2003 Ethiopia 2000

Ethiopia: 2000-2005 (Reduced A more than H)

Ghana 2008 Nigeria 2008 Ethiopia 2005 Ghana 2003 Nigeria 2003 Ethiopia 2000

Nigeria 2003-2008 (Reduced H more than A)

Ghana 2008 Nigeria 2008 Ethiopia 2005 Ghana 2003 Nigeria 2003 Ethiopia 2000

Ghana 2003-2008 (Reduced A and H Uniformly)

Ghana 2008 Nigeria 2008 Ethiopia 2005 Ghana 2003 Nigeria 2003 Ethiopia 2000

Pathways to Poverty Reduction

• 6
• 5
• 4
• 3
• 2
• 1

Ghana Nigeria Ethiopia

Annualized Absolute Change in the Percentage Who is Poor and Deprived in... Assets Cooking Fuel Flooring Safe Drinking Water Improved Sanitation Electricity Nutrition Child Mortality School Attendance Years of Schooling

0.1 0.2 0.3 0.4 0.5 0.6 0.7

Nigeria: Indicator Standard Errors

Ethiopia’s Regional Changes Over Time

Addis Ababa Harari

Nigeria’s Regional Changes Over Time

South South North Central

Inside the Regions of Nigeria

• 8.0
• 7.0
• 6.0
• 5.0
• 4.0
• 3.0
• 2.0
• 1.0

0.0 1.0 2.0 3.0

North Central North East North West South East South South South West

Annualized Absolute Change in the Percentage Who is Poor and Deprived in...

Assets Cooking Fuel Flooring Safe Drinking Water Improved Sanitation Electricity Nutrition Child Mortality Years of Schooling School Attendance

Robustness checks to ‘value judgements’ (choice of parameters)

Robustness to poverty cutoff k= 20% to 40%

• 90% of the possible pairs of countries have a

dominance relation for k 2 to 4. That is, we can say that one country is unambiguously poorer than another regardless of whether we require to be poor in 20, 30 or 40% of the weighted indicators.

Robustness to Weights

MPI Weights 1 MPI Weights 2 MPI Weights 3 Equal weights: 33% each (Selected Measure) 50% Education 25% Health 25% LS 50% Health 25% Education 25% LS Pearson 0.992 Spearman 0.979 Kendall (Taub) 0.893 Pearson 0.995 0.984 Spearman 0.987 0.954 Kendall (Taub) 0.918 0.829 Pearson 0.987 0.965 0.975 Spearman 0.985 0.973 0.968 Kendall (Taub) 0.904 0.863 0.854 Number of countries: 109 MPI Weights 2 50% Education 25% Health 25% LS MPI Weights 3 50% Health 25% Education 25% LS MPI Weights 4 50% LS 25% Education 25% Health

Robustness to Weights

Summary:

• High Correlations: 0.97 and above
• High Rank Concordance: 0.90 and

above

• 85% of all possible pairwise

comparisons are robust

• Complements monetary poverty measures
• Gives a ‘high resolution’ lens on poor people’s lives
• An overview and a ‘dashboard’
• Changes over time – can change relatively quickly
• Provides incentives to reduce intensity and incidence.
• Can be used to identify the poorest
• Adaptable for National Poverty Measures (or M&E)
• Research and Policy: large agenda is ongoing

Uses of an MPI

Value Judgement 1: Selection of Data

• existing data
• internationally comparable
• legitimacy: MDGs
• malnutrition rather than income
• updated infrequently

Value Judgement 2: Selection of Dimensions

Dimensions are ‘notional’ – affect wts, communication Follows precedent: HDI, HPI Possible because of data

Value Judgement 3: Selection of Indicators

Data constrained Reflect MDGs (consensus) Relatively Comparable *note: technicalities in construction are hardly normatively justified but matter a lot – e.g. hh

Value Judgement 4: Selection of Weights

Weights are ‘nested’ – equal then equal. This is easy to understand, and often used. Robust to a range of plausible weights Most controversial. 1/6 on Health and Education Indicators 1/18 on standard

• f living indicators.

Identification: Who is poor?

A person is multidimensionally poor if they are deprived in 33% of the dimensions at the same time.

33%

Colombia’s National MPI: 5 Dimensions, 15 Variables, Nested Weights

Educational Conditions Childhood & Youth

Work Health

Housing & Public Services

Schooling Illiteracy School Attendance At the right level Access to infant services No Child Labour Absence of long-term unemploy- ment Coverage Access to health care given a necessity

Improved Water

Flooring Overcrowding Sanitation Exterior Walls Formal work 0.1

0.2 0.2 0.2 0.2 0.2

0.05 0.1 0.1 0.04

Poverty cutoff = 33%

Territorial

Mexico’s National Measure:

6 social deprivations (1/12) + income (1/2)

Social Rights

Deprivations

Population Wellbeing

Income

Current income per capita Six Social Rights:

• Education
• Health
• Social Security
• Housing
• Basic Services
• Food

3 2 1 4 5 6

Social Rights Deprivations

Mexico’s Identification:

poverty = (income + 1); extreme = (lower income + 3) With Deprivations

EXTREME Multidimensional Poverty

3

Moderate Multidimensional Poverty

Vulnerable by social deprivations

Vulnerable by income

5 2 4 1 6

Ideal Situation

Without

D e p r i v a t i

• n

s

MULTIDIMENSIONALLY POOR

Basic Needs £ Food £

Income

102

Day 1: Purpose of poverty measure Dimensions to communicate measure Particular Indicators used to measure poverty Deprivation Cutoffs how much of each indicator? Poverty Cutoff who is poor? Values/Wts what are the relative weights of indicators? Day 2: Procedure: Who decides normative issues? What is the appropriate role of poor people, governments, and statistical or technical experts? Plural Criteria: How should statistical, political, and participatory input be coordinated in measurement design?

Thank you

www.ophi.org.uk

Purpose

105

Aim of each session - discuss how to:

• make the value judgements inherent in this decision (options)
• balance normative, technical, & political issues (priority)
• update over time

Chair to summarize:

• 1. How would we suggest to those charged with designing the multidimensional

poverty measure that they undertake the three steps above?

• 2. What pressing research questions have been noted?
• 3. (does the question need to be reformulated or changed?)

Participants to contribute:

• Literature – please jot or email annotated biblio, stating why you propose each

and what you see it adding (ophi@qeh.ox.ac.uk)

• People / Projects with expertise or research (as above)
• Ideas that you feel like sharing in writing

106

Purpose of poverty measure

For today, we presume that the value judgements pertain to the design of a long-term official measure of multidimensional poverty. The poverty measure will inform policy design, and reflect positive change that can be influenced by public policy. This is to be updated periodically (say every 2 years) using time series data that are nationally representative and can be decomposed by region and relevant social groups. The survey design will take place after the measure is designed.

107

Purpose of poverty measure

The purpose of the evaluative exercise shapes all choices, e.g. National Poverty Measure – to span decades Youth Poverty Measure – once, to profile youth issues Targeting exercise – to benefit poorest of the poor Monitoring measure – to track progress to given goals International Comparisons – across nations Community Development – show changes transparently

108

Purpose of poverty measure

To some extent, the purpose, having been determined, shapes the value judgements. Should these be taken ‘as given’?

E.g. a measure designed to monitor progress towards a national development plan might systematically exclude public debate. Should omission of public debate require justification? Space of resources? E.g. a measure designed to document a given set of human rights from the universal declaration might ignore cultural values. How justify the ‘need’ for contextual vs comparable measures ? E.g. a very rigorous measure designed to evaluate a small poverty intervention may cost more than the intervention itself. E.g. a measure run in a famine-prone area may be framed to exclude malnutrition E.g. a measure may be designed to target 20% of people when 50% are destitute

Other early questions for a measure

• 1. Legal basis? (how endure across time)
• 2. How to update – Data / Survey; Frequency
• 3. Who will update (Institution; authority)
• 4. What Incentives it provides (ministries)
• 5. Political process of developing measure.
• 1. Public Consultations?
• 2. Expert Group – National Statistics &

Economics

• 3. International/Regional Experts?

Dimensions

111

Purpose of poverty measure

For today, we presume that the value judgements pertain to the design of a long-term official measure of multidimensional poverty. The poverty measure will inform policy design, and reflect positive change that can be influenced by public policy. This is to be updated periodically (say every 2 years) using time series data that are nationally representative and can be decomposed by region and relevant social groups. The survey design will take place after the measure is designed.

112

Dimensions to communicate measure

Assume: Dimensions are to be articulated in the space of capabilities and functionings Recall: Dimensions are conceptual categories. They do not appear in the ‘matrix’. Each indicator belongs to one dimension. They often help to set (nested) weights

Colombia’s National MPI: Dimensions emerge from National Plan

Educational Conditions Childhood & Youth

Work Health

Housing & Public Services

Schooling Illiteracy School Attendance At the right level Access to infant services No Child Labour Absence of long-term unemploy- ment Coverage Access to health care given a necessity

Improved Water

Flooring Overcrowding Sanitation Exterior Walls Formal work 0.1

0.2 0.2 0.2 0.2 0.2

0.05 0.1 0.1 0.04

Poverty cutoff = 33%

Territorial

Mexico’s National Measure:

Dimensions named by law

Social Rights

Deprivations

Population Wellbeing

Income

Current income per capita Six Social Rights:

• Education
• Health
• Social Security
• Housing
• Basic Services
• Food

3 2 1 4 5 6

115

Dimensions to communicate measure

Observations:

• Fits into ‘selection of capabilities’ (as broad categories)
• via public debate, consensus instruments, researcher choice,

legal/institutional mandates, empirical studies, data (MPI)

• Clearest issue in theory & practice

Concerns:

• Priority: Don’t other value judgements matter ‘more’?
• Timing: planning ex ante vs response to a clear measure

Practical Issues:

• How update?
• How combine?
• How document?

Indicators

? more powerful; least discussed ?

117

Indicators – More powerful than dimensions?

Constraints: Need to come from some data source (functionings?) Finance & politics constrains content, periodicity, quality Some Considerations are not purely normative:

• data exist or could exist;
• stock vs. flow
• individual vs. household vs cty
• comparability across all ages/ethnicities
• higher quality vs lower quality indicators (£ & survey)
• statistical associations across indicators
• can be changed by public policy

• Frequent usage (national or international);

literature review; discussion with experts;

• ther indicators. IPM-OPHI Internacional,

NBI, ICV y Sisbén III.

• 1. Indicators can be affected by public

policies.

• 2. Availability of information (in the survey of

Quality of Life in Colombia).

Precision of the sample to estimate the variable – estimated coeff of variation <15%.

*EL DANE utiliza: 0-7: Estimación precisa 8-14: precisión aceptable 15-20 ó 15-25: Precisión regular y por lo tanto se debe utilizar con precaución

Selection of Indicators (Variables) Colombia’s MPI

Criteria for variable selection Criteria to validate variables

Mexico’s Multidimensional Poverty Presentations report the 6 dimensions, and do not share indicators or cutoffs. Do indicators matter - other than for experts?

Justification of Indicators

• Links to and proxy the dimensions/capabilities

– E.g. water. health/asset/dimension/gender – E.g. indicators for health capabilities? – In practice, rarely discussed; rarely debated. – Arguably indicators of functionings (BMI, Ed) or their proxies

• Technical issues often rule:

– Accuracy, measurement error, expense, non-response – Tracks changes in poverty over place and time – Large debates even when clear analysis: stunting vs undernutrition

120

Justification of Indicators

• How frame the problem & debate?

– Conceptual categorization? (e.g. water) – Best proxy for definition of capability/dimension? – Choice of priority among technical criteria? – Take actual issues one by one (e.g. time use) – Is normative input ‘essential’ vs ‘possibly helpful’ – Should the ‘choice of dimensions’ become ‘choice of indicators?’

• But too technical for public debate?

121

Cutoffs

z ( 13 12 3 1) Cutoffs ฀ g0  1 1 1 1 1 1 1               2 4 1

Recall: two cutoffs. Both clearly vj.

Deprivation Cutoffs zj: if a person is deprived in each indicator Poverty cutoff k, a person is poor if ci > k (Ex: k = 2) Indicators Indicators c Persons ฀ y  13.1 14 4 1 15.2 7 5 12.5 10 1 20 11 3 1              

Deprivation Cutoffs zj: if a person is deprived in each indicator Poverty cutoff k, a person is poor if ci > k (Ex: k = 2) Indicators Indicators c (k) Persons z ( 13 12 3 1) Cutoffs

4 2 1 1 1 1 1 1               g

Recall: two cutoffs

฀ y  13.1 14 4 1 15.2 7 5 12.5 10 1 20 11 3 1              

Deprivation Cutoffs:

Clearly are value judgements: How much is enough not to be deprived? – Example: Income Poverty Line – Example: MPI – MDG indicators Clearly matter fundamentally:

• Affect ‘effective weights’
• Define possibility to be identified as poor
• Empirically, can be greater sensitivity

Justification of deprivation cutoffs

• Technical (although disputed)

– E.g. safe water. Particular bugs absent (response codes) – E.g. malnutrition. Z scores and reference groups – Statistical properties

• Political & Legal

– Promised / Required (e.g. compulsory education, plan)

• Constraints & Challenges:

– Diversity – individual & group – Knowledge of data concerns & analyses – Knowledge of possibilities – Comparability (rural-urban; climatic zones)

126

Poverty Cutoffs:

Clearly a value judgment:

How much is enough to be poor? – Reflects purpose (targeting vs national measure) – Often political interest This is a new step – so not many precedents. Has been set

• To match particular headcount ratio
• To reflect participatory or subjective assessments
• To match legal definition (Mexico)
• To match statistical ‘gaps’ in data points (Bristol)

The number of MPI deprivations experienced by those who were income poor, and those who perceived themselves to be poor, was compared with the number

• f deprivations among the non-income and non-subjective poor.

Poverty Cutoff – Colombia.

Median Average People who perceive themselves to be poor 5.0 5.0 Income poor people 5.1 5.2 Income poor people who perceive self as poor 5.4 5.6 Those who don’t perceive themselves as poor 3.0 3.2 Those who are not income poor 3.0 3.2 All people 3.8 4.1

Median and Average number of deprivations 2008

Fuente: Cálculos DNP-SPSCV, con datos de la ECV2008

A non-poor person on average has 3 deprivations, which suggests that a low value of k would capture deprivations that were not related to or sufficient to identify poverty.

Social Rights Deprivations

Mexico’s Poverty Cutoffs:

poverty = (income + 1); extreme = (lower income + 3) With Deprivations

EXTREME Multidimensional Poverty

3

Moderate Multidimensional Poverty

Vulnerable by social deprivations

Vulnerable by income

5 2 4 1 6

Ideal Situation

Without

D e p r i v a t i

• n

s

MULTIDIMENSIONALLY POOR

Basic Needs £ Food £

Income

Weights

Weights (Values)

• Some critics have focused on the weights

– Claiming they cannot be set in a defensible way – Claiming disputes on weights undermine legitimacy of measure – Prefer a ‘mechanical’ route – eigen vectors/regression coefficients/prices

• Thus far, insufficient guidance

– Yes, weights are normative, and not embarrassing to set – Yes, we will disagree hence need a plausible range of weights, but: – Weights are also a function of deprivation cutoffs / headcounts – Weights are also influenced by association among indicators – Weights vary across person/group: combine or apply individually?

131

Weights (Values)

• Literature is large (review paper)

– Is it sufficient to explain the options, strengths & weaknesses? – Is there anything that can be ruled out

• Thus far, insufficient guidance

– Yes, weights are normative, and not embarrassing to set – Yes, we will disagree hence need a plausible range of weights, but: – Weights are also a function of deprivation cutoffs / headcounts – Weights are also influenced by association among indicators – Weights vary across person/group: combine or apply individually?

132

Equal (nested) weights

• Most commonly used approach
• Advocated for policy communication

(Atkinson)

• Equal weights represent value judgments
• Example:
• 1. BMI, years of school (0.5)
• 2. BMI, yrs school, caloric intake, anaemia, (0.25)

Field Studies: Participatory FGD

–The Participants:

–Identified the focal problems of poverty –Ranked the dimensions of poverty (weights) –Identified ‘cutoffs’ – who is poor? –Provided feedback on the 3 trial measures

135

Another community: FGD

Ruepisa: Ranking Most important

Electricity Land Sanitation Health Drinking Water

Next most

Education Housing

Third

Income / Money

Fourth

Animal Assets