OPHI Oxford Poverty & Human Development Initiative www.ophi.org.uk

Multidimensional Poverty Index Sabina Alkire and Maria Emma Santos, July 2010

Key Findings Research Brief

■ Half of the world’s MPI poor people live in South Asia; and just over a The Oxford Poverty and Human Development Initiative (OPHI) quarter in Africa. has developed a new international measure of poverty – the ■ People living in MPI poverty may Multidimensional Poverty Index or MPI – for the 20th Anniversary not be income poor. For example, edition of the United Nations Development Programme’s flagship only two-thirds of Niger’s people are income poor, whereas 93% are Human Development Report. The new innovative index goes beyond poor by the MPI. a traditional focus on income to reflect the multiple deprivations

■ Multidimensional poverty varies that a poor person faces with respect to education, health and living a lot within countries. In Kerala, standard. This brief summarises the new method and key findings and India 16% of people are MPI poor; shows how the MPI can be used. compared to 81% in the Indian state of Bihar. The MPI assesses the nature and Income intensity of poverty at the individual What does it mean to live in ■ The composition of poverty level, with poor people being those poverty? This question has often varies. We found five ‘types’ of multidimensional poverty among who are multiply deprived and the been answered by lack of income, the countries, each of which extent of their poverty being measured but the traditional narrow focus on require different policy responses. by the extent of their deprivations. The income as the only measure of a MPI creates a vivid picture of people person’s wellbeing, or lack of it, is ■ Ethiopia reduced MPI poverty living in poverty within and across being increasingly challenged. Recent by improving nutrition and water, countries, regions and the world. high profile initiatives, such as the whereas Bangladesh improved It is the first international measure Stiglitz-Sen-Fitoussi Commission, it by sending children to school. Ghana improved several aspects of its kind, and offers an essential have called for broader measures that of MPI poverty at once. complement to income poverty take account of other vitally important measures because it measures aspects of life. ■ Although limited by data, the MPI deprivations directly. The MPI can be The human development approach is robust. 95% of the rankings do used as an analytical tool to identify has long argued that although not change if we look at people the most vulnerable people, show income is important, it needs to who are poor in as little as 20% or aspects in which they are deprived be complemented by more direct as many as 40% of deprivations. and help to reveal the interconnections measures (Anand and Sen 1997). among deprivations. This can enable In 2010, the 20th anniversary year policy makers to target resources and of the United Nations Development design policies more effectively. Programme’s flagship Human Oxford Poverty & Human OPHI has just concluded a first Development Report (UNDP HDR), Development Initiative (OPHI) ever estimate and analyses of global the HDR is introducing a new Department of International Development multidimensional poverty in 104 international measure of poverty – the Queen Elizabeth House (QEH) developing nations across the world Multidimensional Poverty Index or University of Oxford, Mansfield Road using existing household survey data. MPI – which directly measures the Oxford OX1 3TB UK Tel: +44 (0)1865 271915 Other dimensions such as work, combination of deprivations that each Fax: +44 (0)1865 281801 safety, and empowerment could be household experiences. The new Email: [email protected] www.ophi.org.uk incorporated into the MPI in the future MPI supplants the Human Poverty as data become available. Index or HPI used in previous Human OPHI gratefully acknowledges support from the This brief summarises key Development Reports. International Development Research Centre (IDRC) in Canada, the Canadian International Development findings of the new index and shows The MPI looks at poverty through Agency (CIDA), Australian Agency for International how the measure can be used by a ‘high-resolution’ lens. By directly Development (AusAID), the UK Department for International Development (DFID), Doris Oliver governments, development agencies, measuring the nature and magnitude Foundation, Global Giving, Kim Samuel-Johnson and other institutions to help eradicate of overlapping deprivations at the and anonymous benefactors. acute poverty. household level, the MPI provides Multidimensional Poverty Index July 2010 information that can help to inform very easy to calculate and interpret, is better policies to reduce acute poverty. Inside the MPI intuitive yet robust, and satisfies many The MPI is the first international 1. Education (each indicator is desirable properties. measure to reflect the intensity of weighted equally at 1/6 ) Better data are needed at the poverty – the number of deprivations • Years of Schooling: deprived if international level to be able to expand that each household faces at the same no household member has com- the measure to include other important time. pleted five years of schooling dimensions, such as informal work, From their inception, the UNDP • Child Enrolment: deprived if any empowerment and safety from school-aged child is not attend- violence, in the future. HDRs have pioneered new ways to ing school in years 1 to 8 analyze human development and By directly measuring the different poverty, intended to have a direct 2. Health (each indicator is weighted types of poverty in each household, impact on development strategy equally at 1/6) the MPI goes beyond the HPI and and methodology. By featuring • Child Mortality: deprived if any other poverty measures to capture child has died in the family this independently conceived new how different • Nutrition: deprived if any adult or approach to poverty measurement in child for whom there is nutritional groups of people the 20th anniversary report, UNDP information is malnourished experience HDR hope to encourage its use by concurrent (each indicator governments, development agencies, 3. Standard of Living deprivations. is weighted equally at 1/18) Take Tabitha, and other institutions dedicated to the • Electricity: deprived if the eradication of acute poverty. household has no electricity who is 44 years It can be decomposed by • Drinking water: deprived if the old and lives in population group and broken down household does not have access Lunga Lunga by dimension. The full results and to clean drinking water or clean slum just outside Tabitha detailed information on the index can water is more than 30 minutes of Kenya’s be found in: Alkire, S. and Santos, walk from home capital, Nairobi. Tabitha lives with her M.E. (2010). Acute Multidimensional • Sanitation: deprived if they do husband and their six children. Her not have an improved toilet or if husband has no permanent work so Poverty: A New Index for Developing their toilet is shared Countries. Oxford Poverty and Human • Flooring: deprived if the house- she is the main breadwinner for the Development Initiative Working Paper hold has dirt, sand or dung floor family. She depends on casual jobs, 38. Available at: http://www.ophi.org. • Cooking Fuel: deprived if they including washing clothes for others in uk/publications/ophi-working-papers/. cook with wood, charcoal or dung the neighbourhood, where she is paid Also see www.ophi.org.uk. • Assets: deprived if the house- Ksh 50 per wash (US $1.65). When no hold does not own more than one one has clothes to be washed, Tabitha Multidimensional Poverty Index of: radio, TV, telephone, bike, or goes to the nearby rubbish dump motorbike, and do not own a car and finds old clothes to sell to a local (MPI): Basic Overview or tractor The MPI is an index of acute clothes recycling dealer who buys 1kg multidimensional poverty. It reflects presented in ‘Inside the MPI’ (above). of clothing for Ksh 10 (US $0.33). On deprivations in education to health The indicators are based on a good day, Tabitha can manage to outcomes to assets and services for participatory exercises with poor collect 1-5kg (total; US $0.33-1.65) of people across 104 countries. Although people, emerging international clothes from the rubbish. deeply constrained by data, the MPI consensus and the availability of Despite having such a low income, reveals a different pattern of poverty suitable data. Most are linked to Tabitha’s four school-aged children than income poverty, as it illuminates Millennium Development Goals. The attend the local school. She has high deprivations directly. The MPI has index mainly uses data from three hopes for her children as neither three dimensions: health, education, household surveys: the Demographic Tabitha nor her husband had the and standard of living. These are and Health Survey, the Multiple opportunity to go to school. “My hopes measured using 10 indicators. Poor Indicators Cluster Survey and the for the future are that I can support my households are identified and an World Health Survey. children to continue their education,” aggregate measure constructed using The MPI is the product of two she says. Tabitha worries that she will a methodology proposed by Alkire numbers: the and Foster (2007, 2009) (See box on Headcount H or page 7). Each dimension is equally percentage of weighted; each indicator within a people who are dimension is also equally-weighted. poor, and the The MPI reveals the combination of average intensity deprivations that batter a household of deprivation at the same time. A household is A – which reflects identified as multidimensionally poor the proportion if and only if it is deprived in some of dimensions in combination of indicators whose which households weighted sum exceeds 30% of all are, on average, deprivations. The indicators and the deprived. Alkire criteria for someone to be considered and Foster show deprived in each indicator are that this measure is Lunga Lunga, Nairobi, Kenya

2 www.ophi.org.uk OPHI Research Brief Multidimensional Poverty Index July 2010

Tabitha at work, looking for old clothes to sell not be able to pay the school fees for secondary level education though. Tabitha’s poverty profile (shaded boxes = poor; non-shaded boxes = non-poor) The family live in a rented house made of iron sheets and a cement their day. (malnutrition, income, drinking water, floor. The house has no toilet, The figure above shows Tabitha’s sanitation); in some cases children electricity or running water. The family poverty profile according to the MPI. (primary school, immunization), or uses one of the public toilets which The shaded boxes show the indicators youth 15-24 (literacy), or childbearing cost Ksh 5 (US $0.16) per visit and in which her household is deprived. women (maternal mortality), or urban buys their drinking water from the Tabitha is poor according to both the slum dwellers (housing) or households’ community water point for Ksh 5 (US MPI and income poverty; the MPI access to secure tenure and so on. $0.16) per jerrican. Because there is tells us more about the nature of the Some environmental indicators do poverty that she faces. not refer to human populations at all. The MPI looks at the poverty of Given this diversity of indicators, it each person in this way. It builds from is difficult to construct an index that the person right up to international meaningfully brings all deprivations level to create a vivid picture of into the same frame. poverty. The index can then be broken The MPI begins to fill this gap. The down by dimension and group to MPI shows which households have clearly show how the composition of key MDG deprivations at the same multidimensional poverty changes in time. Eight of the MPI’s ten indicators incidence and intensity for different relate to MDG targets. Hence the regions, countries, states, ethnic MPI can be used to identify the most groups and more. vulnerable people and identify different Washing clothes at home patterns of deprivations – clusters of no electricity, Tabitha works to daylight The MPI and the MDGs deprivations that are common among hours; she gets up early to prepare her The Millennium Development different countries or groups. The breakfast for her family, if there is food Goals (MDGs) are the most broadly MPI can be used to understand the available. Going without meals is a supported, comprehensive and interconnections among deprivations, weekly occurrence for Tabitha’s family, specific development goals the world help target aid more effectively to but because it happens often, she has ever agreed upon. the most vulnerable, identify poverty talks about it in a light-hearted way: Adoption of the MDGs has traps and consequently strengthen “Going without meals; this is normal increased comparable international the impact of interventions required to for us”. For the evening meal Tabitha data related to the goals and targets, meet the MDGs. prepares provided feedback on development Recent research has shown that Ugali (made outcomes and created incentives to a greater understanding of these from boiled address core deprivations. Unlike the interconnections is key to policy water and MPI, however, the international MDG success. In June 2010, the UNDP maize flour) reports invariably present progress on released an assessment of What it on charcoal. each indicator singly. No composite would take to reach the Millennium Having no MDG index has been developed, Development Goals using detailed TV or radio and few studies have reflected the studies in 50 countries. The first for enter- interconnections between indicators. key message is that the MDGs tainment, There are two reasons that no are interconnected so we need to the family composite MDG index has been address MDG deprivations together. spends their developed. First, the data often come “Acceleration in one goal often speeds evenings from different surveys. Second, up progress in others…Given these sitting even when the data are in the same synergistic and multiplier effects, together as survey, the ‘denominator’ or base all goals need to be …achieved a family and population of MDG indicators differ. simultaneously.” talking about De-threading old clothes In some cases it includes all people Finally, a note on reporting

3 www.ophi.org.uk OPHI Research Brief Multidimensional Poverty Index July 2010 conventions. Many MDG reports focus Arab States on the percentage of countries that are 2% Central and Eastern ‘on target’ to meet the MDGs. which Europe and the CIS under-emphasizes poor people in 1% large countries. Our analysis using the MPI emphasizes the number of poor people whose lives are diminished by multiple deprivations. East Asia and the Pacific 15% Initial Findings Using the MPI Latin America and OPHI analysed data from 104 the Caribbean countries with a combined population 3% of 5.2 billion (78 per cent of the world total) using the MPI (Alkire and Santos 2010). The results should South Asia be considered the first analysis of 51% multidimensional poverty rather than a comprehensive ranking. The 2010 MPI Sub-Saharan Africa was created to reflect poverty in less 28% developed countries. Key findings of the analyses are summarised below.

About 1.7 billion people in the countries covered – a third of their entire population – live in multidimensional poverty, according to the MPI. This exceeds the number of people in those countries estimated to live on US Figure 1: Regional Distribution of People Living in MPI Poverty $1.25 a day or less (1.3 billion), the World Bank’s measure of ‘extreme’ Half of the world’s poor as West Bengal live in multidimensional income poverty. It is less than the total measured by the MPI live in South poverty; the 26 poorest African number of people living on less than Asia (51 per cent or 844 million countries are home to 410 million MPI US $2 a day. people) and over one quarter in poor people. India has experienced Africa (28 per cent or 458 million) strong economic growth in recent The MPI also captures distinct (Figure 1). years, yet the MPI reveals that and broader aspects of poverty. The extensive acute multidimensional percentage of people living in poverty The incidence of MPI poverty is poverty persists. according to the MPI is higher than greatest in Sub-Saharan Africa and the percentage living on less than US South Asia. In Sub-Saharan Africa $2 a day in 43 countries and lower 64.5 per cent of people are MPI than those living on less than US poor; in South Asia, 55 per cent. $1.25 a day in 25 countries. In some Map of MPI Poverty in India countries, the difference between The intensity of poverty – the (higher MPI value in dark red) MPI poverty and income poverty is average number of deprivations particularly marked. For example, in experienced by each household Ethiopia 90 per cent of people are – is also greatest in Sub- MPI poor compared to 39 per cent Saharan Africa and South extreme income poor, and in Pakistan Asia. The poorest country as 51 per cent are MPI poor compared measured by the MPI, Niger, to 23 per cent extreme income poor. also has the greatest intensity Conversely, in Tanzania 89 per cent of poverty (where 93 per cent of people live in poverty and are are extreme income poor compared deprived across 69 per cent of the to 65 per cent MPI poor. The MPI indicators on average). captures deprivations directly – in health and educational outcomes and There are more MPI poor key services such as water, sanitation people in eight Indian states and electricity. In some countries, than in the 26 poorest African these resources are provided free or at countries combined 421 million low cost; in others, it is very hard even people in the Indian States of for working people with an income to Bihar, Chhattisgarh, Jharkhand, obtain them. Madhya Pradesh, Orissa, Rajasthan, Uttar Pradesh, and

4 www.ophi.org.uk OPHI Research Brief Multidimensional Poverty Index July 2010 0.0 The MPI reveals great variation Mexico in poverty within countries. The China Brazil Central Nairobi capital city of Kenya, Nairobi, has the Dominican Republic Urban same MPI value as the Dominican Indonesia Republic, which ranks in the middle 0.1 of the countries analysed, whereas rural areas of northeastern Kenya Ghana Bolivia have a worse MPI value than Niger, Central the poorest of all countries analysed 0.2 (Figure 2). Central Rural

The composition of poverty India differs among regions and ethnic 0.3 Eastern groups. For example, different lu e Kenya Western Coast ethnic groups in Kenya with similar Rift Valley rates of poverty experience different Tanzania Nyanza

deprivations. Deprivation in child MPI Va 0.4 mortality and malnutrition (both health indicators) contribute most to the poverty of the Kikuyu (39 per cent Mozambique of whom are MPI poor), whereas 0.5 deprivations in living standard, such North Eastern as access to electricity, adequate Urban sanitation and cooking fuel, contribute Mali most to the poverty of the Embu 0.6 (37 per cent of whom are MPI poor) (Figure 3). Decomposition of poverty Niger in India and Bolivia also reveals North Eastern interesting differences among groups. 0.7 North Eastern Rural

0.8 Figure 2: MPI Poverty Varies Widely Across Regions of Kenya (above) Figure 3: Comparison of the Percentage of People Who are Poor and Deprived in Each Indicator in two Ethnic Groups in Kenya (below) 15%

12%

9%

Kikuyu 6% Embu

3%

0% Years of Child Child Nutrition Electricity Sanitation Water Floor Cooking Asset Schooling Enrolment Mortality Fuel Ownership

5 www.ophi.org.uk OPHI Research Brief Multidimensional Poverty Index July 2010

Figure 4: Bubble Chart Showing Relationship Between the Percentage of MPI Poor People, Average Intensity of MPI Poverty and Income (above). Low income countries are spread across the chart, from Uzbekistan to Niger. Countries with greatest MPI poverty (highest incidence and greatest in- tensity) are located in the top right of the chart.

Figure 5: The MPI Reveals Five Distinct Types of Deprivation Across Countries: Percentage Contribution of Each Dimension to the Overall MPI Poverty in Each Group (below)

Asset Ownership 100%

Cooking Fuel 90%

80% Floor

70% Water

60% Sanitation 50% Electricity 40%

Nutrition 30%

20% Under 5 Child Mortality

10% Child Enrolment

0% Group 1 Group 2 Group 3 Group 4 Group 5 Years of Schooling

6 www.ophi.org.uk OPHI Research Brief Multidimensional Poverty Index July 2010 Multidimensional poverty varies water, whereas Ghana improved • Reflect the results of effective widely among low GDP per capita several indicators at once. policy interventions quickly. The countries. The percentage of people MPI can be quicker to reflect the in low income countries who are The MPI as a Tool for Policy Makers effects of changes in policies MPI poor ranges from 2 percent in Not only is the MPI a more multi- than income alone, where the Uzbekistan to 93 percent in Niger. faceted and more accurate tool for indicators include direct and Figure 4 shows the percentage of measuring poverty, it can also be used dynamic outcomes. people living in poverty in each of the as a tool for eradicating poverty. How • Integrate many different aspects 104 countries analysed, against the can it improve our current toolbox? of poverty related to the MDGs average intensity of their poverty. The To empower poor people to move into a single measure, reflecting size of each bubble represents each out of poverty it is important to look interconnections among country’s population size. Countries holistically at key components of deprivations and helping to identify with a low GDP income (dark red) are poverty – nutrition, years of schooling, poverty traps. spread across the whole chart. The adequate sanitation, clean water, chart also tells the sad tale that in etc. – as all part of one dynamic. Until countries with the highest incidence of now, many of these aspects have The Alkire Foster Method poverty, such as Niger and Ethiopia, been measured in isolation, such as poverty is most intense. the MDGs. The MPI integrates them An Innovative Technique for Mul- into a single measure that can be tidimensional Measurement Analysis of patterns of broken down by geographic area and deprivation across countries population group and analysed to OPHI created the Multidimensional reveals five types of explore how deprivations interconnect. Poverty Index using a technique multidimensional poverty (Figure The 2010 Millennium Goals developed by Sabina Alkire, OPHI Director, and James Foster, OPHI 5). The structure of deprivation Report stressed that the MDGs will Research Associate and Professor of varies. And these variations need be fully achieved only by focusing Economics and International Affairs at to inform policy. For example, some increased attention to those most George Washington University. The countries such as Syria, Iraq and vulnerable and by introducing policies Alkire Foster method measures out- Azerbaijan (Group 5 in Figure 5), and interventions that eliminate comes at the individual level (person are more deprived in health and the persistent or even increasing or household) against multiple criteria education than they are in living inequalities between the rich and the (dimensions and indicators). standard. Countries such as India poor, between those living in rural and Bangladesh experience higher or remote areas or in slums versus The method is flexible and can be used with different dimensions, indica- deprivation in nutrition than in child better-off urban populations, and those tors, weights and cutoffs to create mortality (Group 2 in Figure 5). Most disadvantaged by geographic location, measures specific to different societ- African countries fall into Group 4. By sex, age, disability or ethnicity. Used ies and situations. For example, the identifying patterns of deprivation, the as an analytical tool, the MPI can method can be applied to measure MPI can help us to understand the help policy makers to identify the poverty or wellbeing, target services interconnections among deprivations, poorest households and groups and or conditional cash transfers and for identify poverty traps and strengthen the different deprivations that they monitoring and evaluation of pro- the impact of policies to reduce face. This can help them to target grammes. poverty in specific aspects, such as aid more effectively to those specific The method can show the incidence, the MDGs. communities. intensity and depth of poverty, as well The MPI goes beyond previous as inequality among the poor, de- Multidimensional poverty can international measures of poverty to: pending on the type of data available change rapidly over time. Trends • Identify the poorest people to create the measure. in the MPI over time show different and aspects in which they are pathways to MPI poverty reduction. deprived. Such information is vital The MPI value is calculated using Bangladesh reduced its MPI to allocate resources where they this method. Mexico used a form of considerably from 2004, when 69 per are likely to be most effective. the Alkire Foster method to create their new national poverty measure cent of people were multidimensionally • Identify which deprivations and Bhutan used it to calculate their poor, to 2007, when multidimensional constitute poverty and which ‘Gross National Happiness Index’. poverty fell to 59 per cent. Bangladesh deprivations are most common improved by sending children to among different groups, so that For more information, see: www.ophi. school. In contrast, Ethiopia reduced policies can be designed to org.uk/policy. poverty by improving nutrition and address their particular needs.

REFERENCES:

Anand, S. and Sen, A. 1997. Concepts of Human Development and Poverty: A Multidimensional Perspective. Human Development Papers. New York: UNDP. Alkire, S. and Foster, J. 2007 and 2009. Counting and Multidimensional Poverty Measurement. OPHI Working Paper 7 and 32. Alkire, S. and Santos, M.E. 2010. Acute Multidimensional Poverty: A New Index for Developing Countries. OPHI Working Paper 38. United Nations Development Programme. 2010. The Millennium Development Goals Report 2010. United Nations Development Programme. 2010. What Will It Take to Achieve the Millennium Development Goals? An International Assess- ment.

7 www.ophi.org.uk OPHI Research Brief Niger Ethiopia Mali Central African Republic Burundi Liberia Burkina Faso Guinea Sierra Leone Rwanda Mozambique Angola Comoros DR Congo Malawi Benin Madagascar Senegal Tanzania Nepal Zambia Nigeria Chad Mauritania Gambia Kenya Contribution Living Standard Bangladesh Haiti Republic of Congo India Contribution Health Cameroon Togo Cambodia Yemen Contribution Education Cote d'Ivoire Pakistan Lesotho Income Under $1.25 per day Lao Swaziland Nicaragua Namibia Bolivia Gabon Honduras Ghana Djibouti Morocco Guatemala Indonesia Peru Tajikistan Mongolia Viet Nam Guyana Paraguay Philippines China Dominican Republic Colombia Brazil Turkey Suriname Estonia Egypt Trinidad and Tobago Azerbaijan Sri Lanka Kyrgyzstan Mexico South Africa Argentina Tunisia Jordan Uzbekistan Armenia Ecuador Moldova Ukraine Macedonia Uruguay Thailand Croatia Russian Federation Albania Bosnia and Herzegovina Georgia Hungary Kazakhstan Multidimensional Poverty Compared with Extreme Income Latvia Belarus Poverty in 104 Developing Nations Czech Republic Slovakia Slovenia

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Percentage of People Living in Poverty

July 2010 • www.ophi.org.uk

The Oxford Poverty and Human Development Initiative (OPHI) is a new economic research centre within the Department of International Development at Oxford University. Oxford PoOPHIverty & Human Development Initiative