What other databases and measurements are out there?

Sabina Alkire, ESCAP, October 2017 The MPI in the Sustainable Development Goals • The Global SDGs, adopted on 25 Sept 2015, address poverty in all its forms and dimensions, opening official space for Multidimensional Poverty Indices.

• The first SDG target (1.1) is to end $1.90/day monetary poverty. The second target (1.2): to halve multidimensional poverty. A number of countries are reporting global MPI.

• MPIs are used: a) to provide a disaggregated picture so as to leave no one behind; b) to inform multi-sectoral policy; c) to see where overlapping inequalities hit the hardest. What other measure? 4 Methodology for the National and Global MPIs

1. Select Indicators, Cutoffs, Values 3. Identify who is poor 4. Use: MPI, Incidence

Intensity & EducationEducation Composition EducationEducation

33% 2. Build a deprivation score for each person MPIs jointly analyse multiple SDGs Example: Global Multidimensional Poverty Index (MPI) MPI: methodology Statistical methods include: Standard errors and confidence intervals for all statistics Statistical inference for all comparisons Validation for component indicators, alone and jointly Robustness tests for cutoffs and weights Axiomatic properties include: Subgroup decomposability and Subgroup consistency Dimensional breakdown, Dimensional monotonicity Ordinality, Symmetry, Scale and replication invariance, Normalization, Poverty and Deprivation Focus, Weak Monotonicity, and Weak Deprivation Re-arrangement

Alkire Foster Seth Santos Roche Ballon OUP 2015 MPIs: National or Comparable Comparable MPIs (Global MPI, ECLAC MPI) - Like $1.90/day and $3.10/day poverty measures - Can also compare countries (& subnational groups, over time) - Could track SDG-1: halve poverty in its many dimensions; - Useful to compare progress and learn from each other. National MPIs: Tailor made for policy - Reflect National Priorities - Compute as official national statistics - Vital for policy: target, coordinate, monitor - Comparable over time, groups, provinces Ecuador Panama

Chile Existing Data – Global MPI Online Global MPI Data:

National National • MPI, H, A, (k=20%, 33%, 50%) • Destitution, Inequality among poor Urban-Rural Disaggregation • Pop, GNI, HDI, $1.90, $3.10, national Subnational Disaggregation poverty, missing indicator • % deprived in each indicator Changes – Harmonised • Uncensored & Censored All MPIs ever published • % contribution of each indicator Age Disaggregation • Standard errors • Sample size Global MPI in ESCAP – 24 countries

Number of countries that are ESCAP members : 53 countries Number ESCAP countries with internationally comparable survey data : 24 countries Number ESCAP countries with comparable subnational data : 15 countries Total number of subnational regions with comparable data : 215 regions Total number of countries with published changes over time : 18 countries Afghanistan Maldives Armenia Myanmar Azerbaijan Nepal Bangladesh Pakistan Bhutan Philippines Cambodia Tajikistan China Thailand Timor-Leste Turkmenistan Kazakhstan Uzbekistan Kyrgyzstan Vanuatu Laos Viet Nam Global MPI in ESCAP – 24 countries - 12 DHS surveys - 9 MICS survey - 1 DHS/MICS - China = CFPS 2014; India = IHDS 2011/12 • 2006 (2) • 2007 (1) 3 countries: 2006-2007 • 2009 (1) • 2009/10 (1)- 6 countries: 2009-2011/12 • 2010 (2) 6 countries: 2012=13/14 • 2011/12 (2) • 2012 (3) 9 countries: 2014-16 • 2012/13 (1) • 2013 (1) • 2013/14 (1) 10 indicators: 19 countries • 2014 (5) 9 indicators: 4 countries • 2015 (1) • 2015/16 (3) 8 indicators: 1 country Global MPI in ESCAP – 3.7 billion people of Kazakhstan 0.001 Armenia 0.001 whom 837 million are poor Turkmenistan 0.001 Kyrgyzstan 0.002 (22.6%) Thailand 0.003 Uzbekistan 0.008 China 0.017 Maldives 0.018 Azerbaijan 0.021 Viet Nam 0.029 Philippines 0.052 Tajikistan 0.054 Indonesia 0.066 Bhutan 0.119 Nepal 0.126 Vanuatu 0.129 Myanmar 0.134 Cambodia 0.146 Laos 0.174 India 0.191 Bangladesh 0.196 Pakistan 0.230 Afghanistan 0.295 Timor-Leste 0.360 0.000 0.050 0.100 0.150 0.200 0.250 0.300 0.350 0.400 Share of MPI poor in ESCAP by major regions

16% MPI poor 0.3% MPI poor

84% MPI poor East Asia and the S Asia Pacific Europe and Central Asia South Asia Multidimensional poverty in 24 ESCAP countries: a comparison between Headcount ratio (H) and Intensity (A) 60%

Pakistan Laos Timor-Leste Afghanistan 50% Philippines Myanmar Bangladesh Bhutan China India Viet Nam Indonesia NepalCambodia 40% Tajikistan Vanuatu Azerbaijan Thailand Kyrgyzstan Uzbekistan Armenia Maldives Turkmenistan Kazakhstan

30% 0 10 20 30 40 50 60 70 80 MPI also varies greatly across subnational regions 75% within a country – e.g. Indonesia

70%

65%

60% Indonesia 55%

50%

45%

40% Average of Intensity Average (A) Poverty 35%

30% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Percentage of People Considered Poor (H) MPI also varies greatly across subnational regions

75% within a country – e.g. Indonesia

70%

65%

60%

55% Indonesia West 50% 45% East Central 40% Average of Intensity Average (A) Poverty 35% DKI DI Yogyakarta 30% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Percentage of People Considered Poor (H) MPI also varies greatly across subnational regions

75% within a country – e.g. Indonesia

70%

65%

60%

55% Indonesia Papua West Papua 50% Maluku East Nusa Tenggara North Sulawesi Southeast Sulawesi 45% North Maluku West Nusa Tenggara Banten 40% Bali Aceh Riau West Java East Kalimantan Average of Intensity Average (A) Poverty 35% DKI Jakarta DI Yogyakarta 30% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Percentage of People Considered Poor (H) Afghanistan 2015/16: MPI H rates vary subnationally

Afghanistan 2015/16 56%

Nepal 2014: 29%

Bhutan 2010: 28%

Pakistan 2012/13 44%

Bangladesh 2014: 41% India 2011/12: 41% Subnational MPI: poorest 24 ESCAP regions

Urozgan (Afghanistan) 0.340 Nooristan (Afghanistan) 0.611 Oecussi (Timor-Leste) 0.508 Ermera (Timor-Leste) 0.497 Badghis (Afghanistan) 0.466 Ainaro (Timor-Leste) 0.444 Kandahar (Afghanistan) 0.437 Viqueque (Timor-Leste) 0.410 Bobonaro (Timor-Leste) 0.406 Balochistan (Pakistan) 0.402 Liquica (Timor-Leste) 0.387 Badakhshan (Afghanistan) 0.387 Ghor (Afghanistan) 0.384 Aileu (Timor-Leste) 0.379 Lautem (Timor-Leste) 0.373 Laghman (Afghanistan) 0.369 Takhar (Afghanistan) 0.361 Saravane (Laos) 0.359 Baucau (Timor-Leste) 0.357 Heart (Afghanistan) 0.353 Manufahi (Timor-Leste) 0.353 Samangan (Afghanistan) 0.350 Sar-E-Pul (Afghanistan) 0.341 Wardak (Afghanistan) 0.340 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50 0.55 0.60 0.65

Myanmar (2016) The composition of MPI can inform policy.

0.250 0.200 Here, Indonesia 2007 (previous period) 0.150 0.100 0.050 0.000

100 Assets 90 Fuel 80

70 Flooring

60 Water 50 Sanitation 40

30 Electricity

20 Mortality 10 Attendance 0 Schooling 0.010 MPI Reduction 2007-12

West Papua

0.005 Papua

North SulawesiCentral Kalimantan West Nusa Tenggara West Sulawesi 0.000 0.00 JakartaEast Kalimantan0.05 South0.10 East Sulawesi 0.15 0.20 0.25 National Reduction Bangka-Belitung North Maluku Yogyakarta Special Aceh in MPI Bali Region Riau -0.005 West Java Kepulauan Bangka- Belitung North East Java -0.010 Maluku East Nusa Tenggara

South Sulawesi Annual Absolute Change in % MPI % in Change Absolute Annual

-0.015 Banten

Multidimensional Poverty Index (MPI) at initial year Good /Good

-0.020

27 0.010 MPI Reduction 2007-12

West Papua

0.005 Papua

North SulawesiCentral Kalimantan Central Sulawesi West Nusa Tenggara West Sulawesi 0.000 Jambi 0.00 JakartaEast Kalimantan0.05 South0.10 East Sulawesi 0.15 0.20 Reduction in 0.25 National Bangka-Belitung North Maluku MPI Yogyakarta Special Aceh Bali South Kalimantan Gorontalo Region Riau -0.005 Central Java West Java Kepulauan Bangka- Lampung Belitung East Java West Kalimantan -0.010 Bengkulu South Sumatra Maluku East Nusa Tenggara

South Sulawesi Annual Absolute Change in % MPI % in Change Absolute Annual

-0.015 West Sumatra Banten

Good /Good

-0.020 Multidimension Poverty Index (MPI) at initial year

28 Nutrition Child Mortality Years of

Schooling

Aceh NorthSumatra West Sumatra Riau Jambi South Sumatra Bengkulu Lampung Bangka-Belitung KepulauanBangka-Belitung Jakarta West Java CentralJava YogyakartaSpecial Region EastJava Banten Bali West Tenggara Nusa East Nusa Tenggara 1 Attendance 0.5 Cooking 0 Fuel Sanitation -0.5 -1 Water -1.5 Electricity

Annualized Absolute Change Absolute Annualized -2 Floor -2.5

-3 Assets in in proportionin... deprived andpoor is who -3.5

29 MPI vs. Income poverty

2

1

0

-1

-2

-3

-4

-5

-6 MPI Incidence $1.25 Incidence -7

-8 Disaggregating by Ethnic Group - Benin

0.010 Multidimensional Poverty Index at initial year

0.005 Size of bubble is proportional to the number of poor in first year of Peulh the comparison. Dendi

0.000

0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 MPI Bétamaribe -0.005 Yoruba Bariba

Fon -0.010

Adja Yoa and Lopka -0.015 Poorest ethnic group saw no -0.020 change in .

-0.025 MPI. Annual Absolute Change in Change Absolute Annual

-0.030

-0.035 Disaggregating by Ethnic Group - Kenya

0.010 Multidimensional Poverty Index at initial year Meru 0.005 Size of bubble is proportional to the number of poor in first Poorest ethnic year of the comparison. 0.000 Kamba Mijikenda/Swahili group reduced MPI 0.00 0.10 Kikuyu 0.20 Luhya0.30 0.40 0.50 0.60 0.70 0.80 MPI the fastest. -0.005 Luo Kalenjin Kisii -0.010

-0.015

-0.020 Somali

-0.025 Annual Absolute Change in Change Absolute Annual

-0.030

-0.035 Absolute Reduction in MPII by Social groups

Muslim () [0.32] Slower progress Hindu (*) [0.306] Religion for STs and Christian () [0.196]

Muslims Sikh (*) [0.115] I I in 1999] ST (*) [0.458] - SC (*) [0.378] Caste OBC (*) [0.301] General (*) [0.229]

Rural (*) [0.368] Urban (*) [0.116]

-0.110 -0.090 -0.070 -0.050 -0.030 -0.010 States (Significance) [MPI 33 Absolute Change (99-06) in MPI-I http://www.ophi.org.uk/multidimensional-poverty-index/mpi-resources/#2016 Across 103 countries and 5.4 billion people Half the MPI poor people are children Leave No One Behind Policy Matters New National MPIs launched as official statistics since September 2015 • El Salvador – MPI based on the ‘protagonists’ of poverty (2015) • Costa Rica – MPI aligns allocation with national goals (2015) • Ecuador –MPI reflects political commitment to Buen Vivir (Feb 2016) • Pakistan –MPI reflects the Vision 2025, in detail (June 2016). • Chile – MPI-2 includes dimension of environment & networks (2016) • Honduras – MPI includes work and informs targeting (August 2016) • Mozambique – MPI shows trends from 1996-2014/15 (Oct 2016) • Armenia – MPI reflects complexity & persistence (November 2016) • Panama – annual MPI profiles high disparity subnationally (June 2017) • Dominican Republic – innovative MPI with digital divide (June 2017)

36 New National MPIs launched as official statistics since September 2015 • El Salvador – MPI based on the ‘protagonists’ of poverty (2015) • Costa Rica – MPI aligns allocation with national goals (2015) • Ecuador –MPI reflects political commitmentMultidimensional to Buen Vivir (Feb 2016) Poverty • Pakistan –MPI reflects the Vision 2025, in detailPeer (June Network 2016). • Chile – MPI-2 includes dimension of environment(www.mppn.org) & networks has(2016) 53 participating countries. • Honduras – MPI includes work and informs targeting (August 2016) • Mozambique – MPI shows trends from 1996The-2014/15 2017 meeting (Oct 2016) is by • Armenia – MPI reflects complexity & persistenceChina (November; the 2018 meeting2016) is • Panama – annual MPI profiles high disparity hostedsubnationally by South(June Africa 2017). • Dominican Republic – innovative MPI with digital divide (June 2017)

37 Policy makers are using national or global MPIs to:

1. Complement monetary poverty statistics 2. Track poverty over time (official statistics) 3. Allocate resources by sector and by region 4. Target marginalized regions, groups, or households 5. Coordinate policy across sectors and subnational levels 6. Adjust policies by what works (measure to manage) 7. Leave No One Behind see the poorest & track trends 8. Be Transparent so all stakeholders engage – NGOs, • Private Sector etc, all parts of government. Pakistan: the poorest district reduced MPI most Starting MPI value vs Absolute Reduction of MPI 2004-2015 15% Ziarat Killa Abdullah 10%

Barkhan 5% D.I. Khan Pishin Chagai Badin Islamabad 0% Chakwal Jaffarabad Kohistan 0.000 0.100 0.200Sheikhupura 0.300 0.400 0.500 0.600 0.700 0.800

-5% Gujrat Zhob

Jhal Magsi -10% Lahore Karachi Sialkot Rajanpur Nasirabad -15% Chitral Rawalpindi Khairpur Buner Kasur Haripur Naushehro Feroze -20% Attock Loralai Jhelum Leave No One Behind Kalat T.T. Singh Malakand -25% Khuzdar Killa Saifullah

Larkana -30% Dadu Musakhel -35% An MPI offers: a Headline, Disaggregation & Interlinkages to inform integrated action to complement monetary measures

40