Lesson 3: The Technical Process of Creating an MPI - Part 1

Designing a Multidimensional Poverty Index

Lesson 3: The Technical Process of Creating an MPI - Part 1

start

Welcome to the course

Designing a Multidimensional Poverty Index

The purpose of this course is to outline the process of designing a national Multidimensional Poverty Index (MPI). This course will illustrate the process of creating a multidimensional poverty measure, describe technical and political processes to create sustainable and rigorous measures that are proactively and effectively used in policy to end poverty, and provide examples based on countries’ experiences.

Click the arrows in the upper right corner to navigate through the course and click the buttons in the timeline below to skip sections.

Step 1: Purpose

Step 7:Poverty cut-off

Step 2:Space

Step 8:Compute

Step 5: Deprivations

Step 4: Dimensions

Overview

Step 3: Identification

AFMethod

Course structure

Welcome

Lesson 3

Lesson 3 completed

Resources/References

Step 6: Weights

Course structure

The lessons will cover the following topics:

Lesson 1: Introduction to the multidimensional approach to poverty eradication

Lesson 1 provided the background for undertaking the multidimensional approach to poverty eradication. It explained the difference between monetary and multidimensional poverty, outlined motivations for the multidimensional approach to poverty eradication, defined a national MPI, and explained its objectives and value.

Step 1: Purpose

Step 7:Poverty cut-off

Step 2:Space

Step 8:Compute

Step 5: Deprivations

Step 4: Dimensions

Overview

Step 3: Identification

AFMethod

Course structure

Welcome

Lesson 3

Lesson 3 completed

Resources/References

Step 6: Weights

Course structure

The lessons will cover the following topics:

Lesson 2: Generating support for national MPIs

Lesson 2 discussed the process of engagement with different actors and how institutional arrangements facilitate the process of designing, computing and using national MPIs. It will also explain the relevance of a solid communications strategy to guarantee the sustainability of the measure over time.

Step 1: Purpose

Step 7:Poverty cut-off

Step 2:Space

Step 8:Compute

Step 5: Deprivations

Step 4: Dimensions

Overview

Step 3: Identification

AFMethod

Course structure

Welcome

Lesson 3

Lesson 3 completed

Resources/References

Step 6: Weights

Course structure

The lessons will cover the following topics:

Lesson 3: The technical process of creating an MPI - Part 1

Lesson 3 presents the Alkire-Foster method and discuss the process of building the multidimensional poverty measure, from the definition of the unit of identification to the selection of the poverty cut-off. The lesson will also present real examples of how countries have made these decisions and provided technical and normative arguments to validate each of them. Finally, the lesson will discuss the different sources of information that can be used when developing a national MPI.

Step 1: Purpose

Step 7:Poverty cut-off

Step 2:Space

Step 8:Compute

Step 5: Deprivations

Step 4: Dimensions

Overview

Step 3: Identification

AFMethod

Course structure

Welcome

Lesson 3

Lesson 3 completed

Resources/References

Step 6: Weights

Course structure

The lessons will cover the following topics:

Lesson 4: The technical process of creating an MPI - Part 2

Lesson 4 will discuss how to analyze candidate measures, how to select the final version of the national MPI and what additional analysis should be conducted. The lesson will also discuss how to analyze changes over time and track progress in the MPI, as well as provide a summary of how the national MPI could be presented to the public.

Step 1: Purpose

Step 7:Poverty cut-off

Step 2:Space

Step 8:Compute

Step 5: Deprivations

Step 4: Dimensions

Overview

Step 3: Identification

AFMethod

Course structure

Welcome

Lesson 3

Lesson 3 completed

Resources/References

Step 6: Weights

Course structure

The lessons will cover the following topics:

Lesson 5: Using national MPIs as policy tools

Lesson 5 will present examples of how countries have used their national MPIs as a policy tool to reduce multidimensional poverty. Naturally, this is essential, because it is only when MPls become a tool for action — when their insights are understood and used to design high-impact policies and activities — that they "move the needle" on poverty. When national MPls and the associated ‘information platform' of indicator details are updated, they monitor trends in SDG Goal 1 and provide a good feedback loop about what deprivations improved and where.

Step 1: Purpose

Step 7:Poverty cut-off

Step 2:Space

Step 8:Compute

Step 5: Deprivations

Step 4: Dimensions

Overview

Step 3: Identification

AFMethod

Course structure

Welcome

Lesson 3

Lesson 3 completed

Resources/References

Step 6: Weights

Lesson 3

Lesson 3: The technical process of creating a national MPI — Part 1

Lesson 3 will provide an introduction to the technical process. lt will present the Alkire-Foster method and discuss the process of building the multidimensional poverty measure, from the definition of the unit of identification to the selection of the poverty cut-off. The lesson will also present real examples of how countries have made these decisions and provide technical and normative arguments to validate each of them. Finally, the lesson will discuss the different sources of information that can be used when developing a national MPI.

Step 1: Purpose

Step 7:Poverty cut-off

Step 2:Space

Step 8:Compute

Step 5: Deprivations

Step 4: Dimensions

Overview

Step 3: Identification

AFMethod

Course structure

Welcome

Lesson 3

Lesson 3 completed

Resources/References

Step 6: Weights

Lesson 3

In Lesson 3, we will:

Describe each step of the design process for a national MPI

objective 1

Showcase real examples of countries' decisions in the design of their national MPI

objective 2

Outline the different sources of information that can be used when developing a national MPI

objective 3

Describe the steps involved in computing an MPI using the Alkire-Foster method

objective 4

Step 1: Purpose

Step 7:Poverty cut-off

Step 2:Space

Step 8:Compute

Step 5: Deprivations

Step 4: Dimensions

Overview

Step 3: Identification

AFMethod

Course structure

Welcome

Lesson 3

Lesson 3 completed

Resources/References

Step 6: Weights

Lesson 3: Glossary

The following terms will be used in Lesson 3. Click the terms on the left side to familiarize yourself with the definitions.

Administrative data

Aggregation

Census

Decomposability

Deprivation cut-off

Dimensions

Flow indicators

Identification

Incidence of poverty

Indicators

Intensity of poverty

Lesson 3: Glossary

The following terms will be used in Lesson 3. Click the terms on the left side to familiarize yourself with the definitions.

Administrative data

Aggregation

Administrative data

Census

Decomposability

Information collected by government agencies or departments, with the main purpose of registering individuals to deliver a service.

Deprivation cut-off

Dimensions

Flow indicators

Identification

Incidence of poverty

Indicators

Intensity of poverty

Lesson 3: Glossary

The following terms will be used in Lesson 3. Click the terms on the left side to familiarize yourself with the definitions.

Administrative data

Aggregation

Aggregation

Census

Decomposability

The mechanism for bringing together the information for each individual into one summarizing statistic.

Deprivation cut-off

Dimensions

Flow indicators

Identification

Incidence of poverty

Indicators

Intensity of poverty

Lesson 3: Glossary

The following terms will be used in Lesson 3. Click the terms on the left side to familiarize yourself with the definitions.

Administrative data

Aggregation

Census

Census

Decomposability

The enumeration of all households and individuals living in a specific territory at a given time.

Deprivation cut-off

Dimensions

Flow indicators

Identification

Incidence of poverty

Indicators

Intensity of poverty

Lesson 3: Glossary

The following terms will be used in Lesson 3. Click the terms on the left side to familiarize yourself with the definitions.

Administrative data

Aggregation

Decomposability

Census

Decomposability

A property of the Alkire Foster measures to be disaggregated by any group for which the data are representative.

Deprivation cut-off

Dimensions

Flow indicators

Identification

Incidence of poverty

Indicators

Intensity of poverty

Lesson 3: Glossary

The following terms will be used in Lesson 3. Click the terms on the left side to familiarize yourself with the definitions.

Administrative data

Aggregation

Deprivation cut-off

Census

Decomposability

The minimum level of achievement that a household or individual must have to be considered non-deprived in each indicator.

Deprivation cut-off

Dimensions

Flow indicators

Identification

Incidence of poverty

Indicators

Intensity of poverty

Lesson 3: Glossary

The following terms will be used in Lesson 3. Click the terms on the left side to familiarize yourself with the definitions.

Administrative data

Aggregation

Dimensions

Census

Decomposability

Conceptual groupings of indicators that are used to communicate the final measure.

Deprivation cut-off

Dimensions

Flow indicators

Identification

Incidence of poverty

Indicators

Intensity of poverty

Lesson 3: Glossary

The following terms will be used in Lesson 3. Click the terms on the left side to familiarize yourself with the definitions.

Administrative data

Aggregation

Flow indicators

Census

Decomposability

Indicators that capture aspects of poverty that are sensitive to changes over time and quickly respond to changes created by social policies or programs (eg the employment rate).

Deprivation cut-off

Dimensions

Flow indicators

Identification

Incidence of poverty

Indicators

Intensity of poverty

Lesson 3: Glossary

The following terms will be used in Lesson 3. Click the terms on the left side to familiarize yourself with the definitions.

Administrative data

Aggregation

Identification

Census

Decomposability

The process of classifying people in the society as poor or non-poor according to certain criteria.

Deprivation cut-off

Dimensions

Flow indicators

Identification

Incidence of poverty

Indicators

Intensity of poverty

Lesson 3: Glossary

The following terms will be used in Lesson 3. Click the terms on the left side to familiarize yourself with the definitions.

Administrative data

Aggregation

Incidence of poverty

Census

Decomposability

The proportion of people identified as multidimensionally poor, also referred to as the "headcount ratio." It is the percentage of people out of the total population whose weighted deprivation score is greater than or equal to the poverty cut-off.

Deprivation cut-off

Dimensions

Flow indicators

Identification

Incidence of poverty

Indicators

Intensity of poverty

Lesson 3: Glossary

The following terms will be used in Lesson 3. Click the terms on the left side to familiarize yourself with the definitions.

Administrative data

Aggregation

Indicators

Census

Decomposability

Fundamental components of the MPI that capture deprivations in functionings that define poverty, according to the purpose of the measure.

Deprivation cut-off

Dimensions

Flow indicators

Identification

Incidence of poverty

Indicators

Intensity of poverty

Lesson 3: Glossary

The following terms will be used in Lesson 3. Click the terms on the left side to familiarize yourself with the definitions.

Administrative data

Aggregation

Intensity of poverty

Census

Decomposability

The average proportion of indicators in which poor people are deprived, or the average deprivation score across all poor people.

Deprivation cut-off

Dimensions

Flow indicators

Identification

Incidence of poverty

Indicators

Intensity of poverty

Lesson 3: Glossary

The following terms will be used in Lesson 3. Click the terms on the left side to familiarize yourself with the definitions.

Intersection approach

Poverty cut-off

Reference population

Space of the measure

Stock indicators

Uncensored headcount ratio

Union identification approach

Units of analysis

Units of identification

Weights

Lesson 3: Glossary

The following terms will be used in Lesson 3. Click the terms on the left side to familiarize yourself with the definitions.

Intersection approach

Poverty cut-off

Intersection approach

Reference population

Space of the measure

An approach to poverty measurement that requires a person to be deprived in all indicators at the same time to be considered multidimensionally poor. That means using a poverty cut—off k equal to 1 or 100 percent.

Stock indicators

Uncensored headcount ratio

Union identification approach

Units of analysis

Units of identification

Weights

Lesson 3: Glossary

The following terms will be used in Lesson 3. Click the terms on the left side to familiarize yourself with the definitions.

Intersection approach

Poverty cut-off

Poverty cut-off

Reference population

Space of the measure

A parameter (k) that identifies those who are multidimensionally poor in at least k weighted indicators. The value of k reflects the minimum level of deprivations or deprivation score an individual or household must be suffering simultaneously to be considered poor.

Stock indicators

Uncensored headcount ratio

Union identification approach

Units of analysis

Units of identification

Weights

Lesson 3: Glossary

The following terms will be used in Lesson 3. Click the terms on the left side to familiarize yourself with the definitions.

Intersection approach

Poverty cut-off

Reference population

Reference population

Space of the measure

The group of people for whom an indicator is relevant and has been effectively measured.

Stock indicators

Uncensored headcount ratio

Union identification approach

Units of analysis

Units of identification

Weights

Lesson 3: Glossary

The following terms will be used in Lesson 3. Click the terms on the left side to familiarize yourself with the definitions.

Intersection approach

Poverty cut-off

Space of the measure

Reference population

Space of the measure

A term that relates to how poverty is measured: space of resources, of inputs, of access to services, of outputs or the space of functionings and capabilities.

Stock indicators

Uncensored headcount ratio

Union identification approach

Units of analysis

Units of identification

Weights

Lesson 3: Glossary

The following terms will be used in Lesson 3. Click the terms on the left side to familiarize yourself with the definitions.

Intersection approach

Poverty cut-off

Stock indicators

Reference population

Space of the measure

Indicators that tend to capture aspects that are stable and are difficult to change with social policies (if at all).

Stock indicators

Uncensored headcount ratio

Union identification approach

Units of analysis

Units of identification

Weights

Lesson 3: Glossary

The following terms will be used in Lesson 3. Click the terms on the left side to familiarize yourself with the definitions.

Intersection approach

Poverty cut-off

Uncensored headcount ratio

Reference population

Space of the measure

The uncensored headcount ratio of an indicator is defined as the proportion of the population that is deprived in that indicator.

Stock indicators

Uncensored headcount ratio

Union identification approach

Units of analysis

Units of identification

Weights

Lesson 3: Glossary

The following terms will be used in Lesson 3. Click the terms on the left side to familiarize yourself with the definitions.

Intersection approach

Poverty cut-off

Union identification approach

Reference population

Space of the measure

An approach to poverty measurement that identifies a person as poor if they are deprived in at least one indicator.

Stock indicators

Uncensored headcount ratio

Union identification approach

Units of analysis

Units of identification

Weights

Lesson 3: Glossary

The following terms will be used in Lesson 3. Click the terms on the left side to familiarize yourself with the definitions.

Intersection approach

Poverty cut-off

Units of analysis

Reference population

Space of the measure

The units for which the Alkire Foster method reflects the joint distribution of disadvantages and analyses poverty (for example, percentage of the population, if the unit of analysis is the population).

Stock indicators

Uncensored headcount ratio

Union identification approach

Units of analysis

Units of identification

Weights

Lesson 3: Glossary

The following terms will be used in Lesson 3. Click the terms on the left side to familiarize yourself with the definitions.

Intersection approach

Poverty cut-off

Units of identification

Reference population

Space of the measure

The units for which the Alkire Foster method reflects the joint distribution of disadvantages and identifies who is poor (for example, if a person is identified as poor on the basis of their and other household members' deprivations).

Stock indicators

Uncensored headcount ratio

Union identification approach

Units of analysis

Units of identification

Weights

Lesson 3: Glossary

The following terms will be used in Lesson 3. Click the terms on the left side to familiarize yourself with the definitions.

Intersection approach

Poverty cut-off

Weights

Reference population

Space of the measure

The value that is given to indicators (and, by association, to dimensions) within the MPI.

Stock indicators

Uncensored headcount ratio

Union identification approach

Units of analysis

Units of identification

Weights

Overview: Steps for designing a national MPI

Click on the images to learn about the steps for designing a national MPI.

Step 4

Step 2

Step 3

Step 1

Select the dimensions and indicators

Select the space of the measure

Select the unit(s) of identification and analysis

Select the purpose(s)

Step 7

Step 5

Step 6

Step 8

Set the deprivation cut-offs for each indicator

Set the weights for each dimension/indicator

Compute the MPI

Set the poverty cut-off

Step 1: Purpose

Step 7:Poverty cut-off

Step 2:Space

Step 8:Compute

Step 5: Deprivations

Step 4: Dimensions

Overview

Step 3: Identification

AFMethod

Course structure

Welcome

Lesson 3

Lesson 3 completed

Resources/References

Step 6: Weights

Step 1

Select the purpose(s)

The design of a national MPI should be guided by the purpose(s) of the measure, keeping the specific context of each country in mind.The purpose(s) of a measure may include its policy applications, the reference population, dimensions, and the frequency of updating and how it will be disaggregated. As discussed in Lesson 1, national MPls can have different purposes depending on the context (for example, monitoring poverty, policy coordination and targeting.) To date, most countries have designed and calculated measures in order to: - Monitor multidimensional poverty - Guide and coordinate public policies - Complement existing income poverty statistics There may be other purposes, such as tracking environmental indicators, the welfare of certain vulnerable groups, or indicators related to peace or freedom.

Step 2

Select the space of the measure

The choice of space determines whether poverty is measured in the space of:

• Resources
• Inputs and access to services
• Outputs
• Functionings and capabilities

Step 3

Select the unit(s) of identification and analysis

These are the units for which the Alkire Foster method reflects the joint distribution of disadvantages, identifies who is poor, and analyses poverty. The unit of identification and analysis might be a person, a household, a region or an institution. They may be the same, but it is not necessary. Poverty measures usually use the individual or the household as the unit of identification, and nearly always use the individual as the unit of analysis.

Step 4

Select the dimensions and indicators

A key step in the development of a national MPI is to decide the structure of the measure: - Dimensions — conceptual categories into which indicators may be arranged for intuition and easy of communication.- Indicators — the building blocks of the measure that bring into view relevant facets of poverty and are used to construct the deprivation scores. - Together, dimensions and indicators measure poverty in the country. - National MPIs are estimated based on indicators (which reflect variables in a dataset), but all countries with existing national MPIs have presented their indicators grouped into dimensions.

Step 5

Set the deprivation cut-offs for each indicator

Setting deprivation cut-offs (the minimum achievement level or category required to be considered non-deprived in that indicator) is a normative exercise. These decisions can be guided by international or national standards (such as the SDGs or national legislation on compulsory education), by the results of participatory or consultative exercises, or by targets included in national development plans. Ultimately, the deprivation cut—offs will reflect the purpose of the measure, data availability, the aspirations of poor persons and communities, the unit of identification and aspects of indicators design discussed in the previous section.

Step 6

Set the weights for each dimension/indicator

Weights, like dimensions, indicators, and cut-offs, are fixed over time. Since by definition weights reflect the value that a deprivation in each indicator has for poverty, relative to deprivations in other indicators, setting weights plays a fundamental role in defining the relative importance of each deprivation in the final measure. The selection of weights reflects normative judgments related to the purpose of the measure. If the objective is to evaluate changes in poverty levels, the weights should aim to reflect the importance of each indicator. On the other hand, if the MPI's purpose is to monitor progress, weights might represent, to some extent, the priorities of the government in reducing deprivations. Most national MPIs use nested weights, with equal weights across each dimension and equal weights across indicators within dimensions, unless there are particular reasons to adjust this.

Step 7

Set the poverty cut-off

Since the poverty cut-off shows what combined share of weighted deprivations is sufficient to identify a person as poor, the setting of this poverty line needs to reflect the priorities and policy goals of the country. For example, if the purpose is to target resources to the poorest populations, the value of k must capture those who are facing the highest number of simultaneous deprivations. In turn, if the goal is to monitor poverty, k should reflect the minimum level of deprivations acceptable in a country to be considered non-poor.

Step 8

Compute the MPI

The M0 or MPI is computed as the product of the incidence of poverty (H) and the intensity of poverty (A). This method not only identifies who is poor but also innovates by incorporating how acute or intense the situation of multidimensional poverty is for the poor. The MPI can be equivalently computed by summing across all indicators each censored headcount ratio multiplied by its respective weight.

Step 1: Purpose of an MPI

Video: Framing the purpose of an MPI

The most common purposes for multidimensional poverty measures include:

• To develop official measures that show the level and composition of poverty, and are regularly updated
• To monitor and evaluate the impact of activities (program specific)
• To compare poverty across regions and groups
• To target the poorest more effectively

Listen to Sabina Alkire’s example of how you can frame the purpose of a national MPI.

Step 1: Purpose

Step 7:Poverty cut-off

Step 2:Space

Step 8:Compute

Step 5: Deprivations

Step 4: Dimensions

Overview

Step 3: Identification

AFMethod

Course structure

Welcome

Lesson 3

Lesson 3 completed

Resources/References

Step 6: Weights

Step 2: Space of an MPI

Video: Creating space for an MPI

Listen to Sabina Alkire elaborate on the considerations for the choice of space for an MPI.

Step 1: Purpose

Step 7:Poverty cut-off

Step 2:Space

Step 8:Compute

Step 5: Deprivations

Step 4: Dimensions

Overview

Step 3: Identification

AFMethod

Course structure

Welcome

Lesson 3

Lesson 3 completed

Resources/References

Step 6: Weights

Step 3: Unit(s) of identification

Individual vs Household as the unit of identification

The unit of identification is the level at which each deprivation is measured. It is the unit at which we identify people or households as poor or non-poor (determining their poverty status). Poverty measures usually use the individual or the household as the unit of identification, and nearly always use the individual as the unit of analysis. The choice is often constrained by data, and depends on the purpose of the MPI. Currently, with the exception of Mexico, all national multidimensional poverty measures use the household as the unit of identification. It is normal for the unit of analysis (how data is reported) to be the individual, even if the unit of identification is the household. That is, one usually reports the percentage of people who are identified as poor rather than the percentage of households, while the unit of identification might remain the household. This is particularly important as poor households tend to have more members. Click on each semi-circle to the right to learn more about the pros and cons of using the household and the individual as units of identification.

Household

Household

Individual

Step 1: Purpose

Step 7:Poverty cut-off

Step 2:Space

Step 8:Compute

Step 5: Deprivations

Step 4: Dimensions

Overview

Step 3: Identification

AFMethod

Course structure

Welcome

Lesson 3

Lesson 3 completed

Resources/References

Step 6: Weights

Key approaches to measuring poverty

Household

With the household as the unit of identification, household members’ information is taken together and combined into household-level deprivations. Thus, all members are equally deprived or non-deprived in each indicator and are equally identified as poor or non-poor. This implies that individual-level indicators like schooling or nutrition are combined across household members. Most national MPIs use the household as the unit of identification. Advantages of using the household unit of identification:

• Addressing the issue of data limitations
• Reflecting the ”sharing and caring” among household members. For example, a person who lives in a household where no one is literate or educated may find themselves in a very different situation than a person who is the only uneducated household member, because other household members can read their letters or bills. Also, the sharing of resources within households tends to be considerable, and households—and not individuals—are the main beneficiary units of many governmental programs.

Limitations:

• Could hide the real level of poverty in the population and be less useful, for example, for public expenditure planning where the precise number of poor people matters.

Key approaches to measuring poverty

Individual

Using the person as the unit of identification means that any individual—level deprivations—for example, in nutrition, schooling or employment— are recorded for each person separately.Advantages of using the individual unit of identification:

• Clear comparisons by gender, age, ethnicity and other relevant individual characteristics
• The ability to analyze intra-household inequalities, such as differences between the levels of education of girls and boys, or employment participation for men and women

Limitations:

• As the MPI requires a complete deprivation profile for each unit, information on all indicators must be available for each person and come from the same source of data. However, most existing data sources used for poverty measurement do not have information for all individuals or for all the indicators usually selected for national MPIs. An alternative, then, is to use the household as the unit of identification and undertake linked gendered and intrahousehold analysis where the underlying microdata are available.

Step 4: Choice of dimensions and indicators

Arguments supporting the choice of dimensions and indicators

Most countries use a combination of multiple criteria to select the most relevant dimensions and indicators to be included into the MPI.

Specific legislation or international conventions Relevant literature and theoretical arguments Goals of national development plans Ongoing deliberative participatory processes and public consultations International or regional examples

Click the arrows to learn more about Step 4.

Step 1: Purpose

Step 7:Poverty cut-off

Step 2:Space

Step 8:Compute

Step 5: Deprivations

Step 4: Dimensions

Overview

Step 3: Identification

AFMethod

Course structure

Welcome

Lesson 3

Lesson 3 completed

Resources/References

Step 6: Weights

Step 4: Choice of dimensions and indicators

Video: Dimensions in Colombia's and Mexico's MPls Example

Listen to Sabina Alkire’s example of the dimensions included in Colombia’s and Mexico‘s national MPls.

Step 1: Purpose

Step 7:Poverty cut-off

Step 2:Space

Step 8:Compute

Step 5: Deprivations

Step 4: Dimensions

Overview

Step 3: Identification

AFMethod

Course structure

Welcome

Lesson 3

Lesson 3 completed

Resources/References

Step 6: Weights

Step 4: Choice of dimensions and indicators

Selecting sources of information

As mentioned earlier, the choice of dimensions and indicators for an MPI is often constrained by data. You will need to decide which data source best allows you to measure poverty based on your measurement goals and your country's particular context. The selection of a data source should be influenced by:1) Normative considerations, and2) Availability of data. Click on the boxes below to learn more.

Normative considerations

Availability of data

Step 1: Purpose

Step 7:Poverty cut-off

Step 2:Space

Step 8:Compute

Step 5: Deprivations

Step 4: Dimensions

Overview

Step 3: Identification

AFMethod

Course structure

Welcome

Lesson 3

Lesson 3 completed

Resources/References

Step 6: Weights

Step 4: Choice of dimensions and indicators

Data sources for dimensions and indicators

Governments can decide if they would like to use data that already exists or new data, depending on the availability of economic and human resources. Click the three options below to learn more.

Existing data source

Modified existing data source

New survey

Step 1: Purpose

Step 7:Poverty cut-off

Step 2:Space

Step 8:Compute

Step 5: Deprivations

Step 4: Dimensions

Overview

Step 3: Identification

AFMethod

Course structure

Welcome

Lesson 3

Lesson 3 completed

Resources/References

Step 6: Weights

Step 4: Choice of dimensions and indicators

Existing data source

The advantages of using an existing data source include:

• Highest feasibility: such a decision reduces the financial and time costs associated with the design and calculation of the national MPI
• Tracking of changes in poverty over time: often an MPI made from existing data can also be back—computed over two or more time periods.

Important considerations to assess whether the use of an existing data source is desirable:

• Quality of data
• Frequency of data collection
• Levels at which data is representative
• Type of information that is covered by existing data sources

The three most common sources of micro-data available are:

• Census
• Household surveys
• Administrative records

Step 4: Choice of dimensions and indicators

Modified existing data source

If a country concludes that no existing survey has all the indicators necessary to compute the national MPI, one option is to modify an existing survey.Important considerations to assess whether modifying an existing survey is desirable:

• The time needed to make these adjustments: Sometimes, the timeframe for presenting the MPI figures is short, meaning that time constraints underpin this decision.
• The associated cost: Some indicators only require including a few questions in the survey, whereas others require a larger number of questions, increasing the time of the survey by 15 minutes or more. Seeking a good balance between informational richness and cost effectiveness requires clarifying such issues early on as policy actors may not know the implications of decisions otherwise.

Having decided to modify an existing survey, the next step is to decide which questions need to be included in the survey (and, by contrast, what will be left out). Relevant stakeholders must engage in a discussion about what dimensions and indicators are fundamental and should be included.

Step 4: Choice of dimensions and indicators

New survey

A new data source can be designed specifically to capture all relevant dimensions and groups that need to be tracked with the national MPI.The advantages of designing a new survey:

• More flexibility-the new survey can be shaped to meet all desired requirements, such as including questions to capture certain innovative dimensions deemed relevant, making improvements to standard questions, being representative at different levels, and including questions that allow for a more detailed analysis of the determinants of poverty.

The disadvantages of designing a new survey:

• It is impossible to back-compute trends over time, and these can be critical to learning what has or has not worked in the past
• Extra cost and resources involved

Step 4: Choice of dimensions and indicators

New survey

Given the latter, most countries decide against creating a new survey.If a country decides to design a new survey, it is important that it is seen as a tool for policy making. It is essential that:

• The purpose of the measure is clear and that there is dialogue with the different stakeholders
• The new data source is representative at the lowest possible level to obtain the most information-rich MPI
• Questions included in the new survey reflect individual or household achievements, depending on the unit of identification chosen
• The data source for a permanent official MPI is sustained, which requires the political and technical commitment to collect the new survey in the future, thus guaranteeing the MPI will be updated in a timely manner.

Step 4: Choice of dimensions and indicators

The universe of indicators

Decisions related to which data source will be used to compute the MPl will affect the possible list of indicators. In all cases, it is important to design a universe of possible indicators. It shows all possible indicators that can be made from the survey that might be relevant to an MPI. There are some aspects of national MPls that are common across countries. Most countries with national MPls have selected a similar set of indicators to include in their national measures, adapting them to their national contexts.

The table above presents a simplified grouping of the list of the dimensions and indicators included in some existing national MPIs. Indicators like school attendance, housing, water and sanitation are nearly universal. Others pertaining to childhood and youth conditions, the environment or social networks are included where relevant.

Step 1: Purpose

Step 7:Poverty cut-off

Step 2:Space

Step 8:Compute

Step 5: Deprivations

Step 4: Dimensions

Overview

Step 3: Identification

AFMethod

Course structure

Welcome

Lesson 3

Lesson 3 completed

Resources/References

Step 6: Weights

Step 4: Choice of dimensions and indicators

Click on the triangles below to learn about the different types of indicators.

Objective/subjective

Outcome/input/ output

Desirable indicators

Flow/stock

Relative/absolute

Step 1: Purpose

Step 7:Poverty cut-off

Step 2:Space

Step 8:Compute

Step 5: Deprivations

Step 4: Dimensions

Overview

Step 3: Identification

AFMethod

Course structure

Welcome

Lesson 3

Lesson 3 completed

Resources/References

Step 6: Weights

Step 4: Choice of dimensions and indicators

Outcome/input/output

An indicator can contain information related to:- the inputs of a process (e.g. number of schools in the area)- the outputs (e.g. school attendance/ years of schooling)- the outcome or final results of an intervention (e.g. knowledge). Any of these three types of indicators can be included in a national MPI, though their selection should be considered in light of the purpose of the measure.

Step 4: Choice of dimensions and indicators

Flow/stock

Flow indicators capture aspects of poverty that are sensitive to changes over time and quickly respond to changes created by social policies or programs (eg. the employment rate). In contrast, stock indicators tend to capture aspects that are stable and are difficult to change with social policies (if at all). Flow indicators are usually preferable for a national MPI because they are better able to guide policy making and show change in response to concrete interventions.

Step 4: Choice of dimensions and indicators

Objective/subjective

Indicators can capture subjective or objective information. In the case of subjective indicators, the aim is to provide information related to the perceptions of individuals about a specific situation (e.g., how they self—assess their health or whether they feel safe in their neighborhoods). Objective indicators are associated with aspects that can be measured directly and are not affected by adaptive preferences (e.g. access to healthcare). Changes in objective indicators over time can be easily interpreted and linked to policy interventions more directly, so they are usually preferred for national MPIs.

Step 4: Choice of dimensions and indicators

Relative/absolute

Relative indicators are defined as relative to a particular society at a particular time. For example, such an indicator is the EU at-risk-of-poverty line defined as 60 per cent of median income in the Member State concerned. Absolute indicators, on the other hand, do not entail the relativity and comparison element. An example is the absolute poverty line defined by the Sustainable Development Goals as $1.90 per day.

Step 4: Choice of dimensions and indicators

Desirable indicators

Different types of indicators can be included in a national MPI. Given that this measure is usually used as a policy tool. you would want indicators that:

• capture changes resulting from policies over time (flow)
• reflect the final impact of a policy (outcome)
• are (objective) measures of those impacts

This being said, it is not always possible to use such indicators, and MPls must sometimes include a mix of indicators within their country measures.

Step 4: Choice of dimensions and indicators

Click on the segments below to learn about the criteria for selecting indicators.

Normative considerations

Statistical validation

Comparability across diverse contexts

Possibility of revision

Avoidance of a large burden for countries

Step 1: Purpose

Step 7:Poverty cut-off

Step 2:Space

Step 8:Compute

Step 5: Deprivations

Step 4: Dimensions

Overview

Step 3: Identification

AFMethod

Course structure

Welcome

Lesson 3

Lesson 3 completed

Resources/References

Step 6: Weights

Step 4: Choice of dimensions and indicators

Normative considerations

Normative considerations are value judgements. They link measurement design back to poor people's lives and values, and forward to the policies that, informed by poverty analysis, will seek to improve their lives. Normative reasoning is about coordinating insights from different analyses and selecting the most optimal options for the design of an MPI. Key sources of insight include:- Deliberative insights (participatory work or documents)- Empirical assessments (data quality, redundancy, robustness) - Expert assessments (stakeholders, historical) - Policy relevance (timing, fit with planned activities)- Practicalities (constraints of data, time, human resources, authority, political will and political feasibility)- Theoretical assessments (legality, human rights)

Step 4: Choice of dimensions and indicators

Statistical validation

All indicators that are presented in the MPI should be validated as being technically sound. This would necessitate the analysis of the quality of each indicator by considering applicable populations, the results of redundancy tests and an analysis of missing values.

Step 4: Choice of dimensions and indicators

Comparability across diverse contexts

Comparability requires an identical definition of deprivations in the different years, including the definitions of the indicators, cut-offs and weights.

Step 4: Choice of dimensions and indicators

Possibility of revision

Institutions responsible for the MPI should participate in a methodological revision, which is recommended once every ten years. The MPI indicators, dimensions, weights and cut-offs will need to be examined to see if they are still the best possible ones to guide policy. New indicators may be required, new data may be available, or even a complete re-thinking of the measurement's structure may be undertaken.

Step 4: Choice of dimensions and indicators

Avoidance of a large burden for countries

The design of an MPI should be deemed feasible for a country, given the resources available.

Step 4: Choice of dimensions and indicators

Click on the icons below to learn about the considerations for indicator design.

Unit of identification and applicable population

Identification or unit-level accuracy

Indicator transformation

Assessing combined measures

Missing values

Step 1: Purpose

Step 7:Poverty cut-off

Step 2:Space

Step 8:Compute

Step 5: Deprivations

Step 4: Dimensions

Overview

Step 3: Identification

AFMethod

Course structure

Welcome

Lesson 3

Lesson 3 completed

Resources/References

Step 6: Weights

Step 4: Choice of dimensions and indicators

Identification or unit-level accuracy

Multidimensional poverty measures require that each indicator accurately identifies each person's or household's deprivations, with the objective that the joint distribution of deprivations is also accurate, on average. In this sense, each indicator should reflect the deprivation faced by each unit of identification across the relevant period and not simply the deprivations faced "on average."

Step 4: Choice of dimensions and indicators

Indicator transformation to match the unit of identification

All indicators included in a national MPI must be calculated using the same unit of identification (individual, household or regions). In cases where the unit of identification is the individual, indicators related to the household (and, in some cases, community) might be included in the measure. For example, Bhutan's Child MPI, calculated using the child as the unit of identification. includes information related to walls and floor material at the household level to evaluate if a child is deprived in terms of living standards (e.g. lives in a household with inadequate walls and floors). The main assumption of this process is that all individuals inside a household are assigned the same level of deprivation in the household-level indicator. In the opposite case, a national MPI using the household as the unit of identification needs to combine information on the education, health and employment achievements of all household members (for whom there is information) into one household—level indicator. For example, a household might be considered deprived if no member has completed sufficient years of schooling.

Step 4: Choice of dimensions and indicators

Unit of identification and applicable population

It is essential to keep in mind what the applicable population is for each indicator included in the MPI. Many early errors in MPI estimation happen because this step is not carefully done. For instance, anthropometric indicators are usually collected for specific groups (such as children under five years of age and women of reproductive age). These indicators are not applicable to other population groups. Similarly, information on employment is only relevant for certain adults, and school attendance is only relevant for children of school-going age. In a given survey, information on these indicators has not been collected for other population groups.

Step 4: Choice of dimensions and indicators

Assessing combined measures

Most existing MPIs are "combined measures" that bring together group-specific deprivations. A special issue needs to be considered when using combined measures—the household composition effect. In particular, when including indicators that refer to deprivations of only a specific population (e.g., children below age five), households with children are more or less likely to be classified as poor depending on their composition. For instance, if all indicators refer to deprivations that are relevant for children, households without children will automatically be identified as non-poor. Of course, this does not mean that the group-specific indicators should not be included in the national MPI, but rather that there should be a balance of relevant populations. In practice you need to see what proportion of households include an 'applicable' person. If it is small — for example if only 5% of households include a child under 2 years, then an indicator on immunization which only applies to children under 2 years of age may not be a good MPI indicator because by definition 95% of person households will be coded as non-deprived.

Step 4: Choice of dimensions and indicators

Missing values

The treatment of missing values is important because mishandling can result in errors. Some indicators with good normative support might have a large number of missing values due to measurement error. Fortunately, the percentage of missing values due to measurement error can easily be identified when computing the levels of deprivation for the applicable population. Remember: a measure based on the Alkire-Foster method can only be created using observations (individuals, households, etc.) that have information on all indicators included in the index. Therefore, this step is fundamental to minimizing the number of observations that are lost.

Go to the next two slides to learn how to deal with missing values.

Step 4: Choice of dimensions and indicators

Missing values

There are two ways of dealing with missing values:

Drop the observations with missing values:

• If the unit of identification is the household, households with a missing value in any MPI indicator are dropped from the sample and not considered when computing the index.

• *If the observations with missing values are systematically different from those with observed values, the reduction in the sample will lead to a bias in the poverty figures, so it is important to assess whether dropping observations with missing values affects the results.

• **In the case of deciding to drop observations with missing values, the reduced sample can still be used if not too many samples are dropped (note: the sampling weights may need to be adjusted). However, this decision should be explicitly mentioned by the researcher, clearly stating whether the poverty estimate is likely to be a "lower” or an "upper" bound, based on the results of a bias analysis. If the two results are not significantly different, then computations can proceed using the reduced sample without affecting the representativeness of the findings.

Step 4: Choice of dimensions and indicators

Missing values

Create a rule to assign a value for the missing data:

• For instance, if there is information for at least 50 percent or 75 percent of household members, then the observation is kept and the values of those household members for whom there is information are used. If there are not enough household members with information to reach that threshold, then the observation is dropped from the calculations.

• *It is imperative to use care when dealing with answers such as "do not know" or "does not answer". Usually these answers are considered missing information, and it is a common mistake to fail to classify them as such.

Step 4: Choice of dimensions and indicators

Integration of SDGs and Environment and Natural Resources (ENR) indicators into a national MPI

The SDGs have several references to environmental degradation and environmental challenges: almost every one of the 17 SDGs relates to the environment and/or poverty.Several technical and political challenges need to be addressed in order to further develop and apply an ENR-MPI at national or global levels:

• Data availability: surveys that are used to calculate the MPI could be amended to include more ENR data.
• Sample design: the criteria according to which the MPI data sampling is stratified may not capture important environmental heterogeneities and be representative of the ENR data.
• Selection of cross-context indicators: the need to select indicators relevant across contexts may pose a greater challenge in the ENR dimension than in others since the environment is very context-specific.

Each of these challenges can be met and provide part of the advocacy agenda for affecting the broader agenda at the national and global levels.

Step 1: Purpose

Step 7:Poverty cut-off

Step 2:Space

Step 8:Compute

Step 5: Deprivations

Step 4: Dimensions

Overview

Step 3: Identification

AFMethod

Course structure

Welcome

Lesson 3

Lesson 3 completed

Resources/References

Step 6: Weights

Step 4: Choice of dimensions and indicators

Click the images below to learn about the options for selecting of indicators.

Design poverty measures for specific groups

Test assumptions about household distribution

Include universally applicable achievements

Step 1: Purpose

Step 7:Poverty cut-off

Step 2:Space

Step 8:Compute

Step 5: Deprivations

Step 4: Dimensions

Overview

Step 3: Identification

AFMethod

Course structure

Welcome

Lesson 3

Lesson 3 completed

Resources/References

Step 6: Weights

Step 4: Choice of dimensions and indicators

Include universally applicable achievements

Include universally applicable achievements

In this case, the measure would only include achievements applicable to the whole population (such as access to a clean source of drinking water or sanitation). Given data limitations, this can substantially reduce the set of possible indicators.

In this case, the measure would only include achievements applicable to the whole population (such as access to a clean source of drinking water or sanitation). Given data limitations, this can substantially reduce the set of possible indicators.

Step 4: Choice of dimensions and indicators

Include universally applicable achievements

Design poverty measures for specific groups

In this case, the measure would only include achievements applicable to the whole population (such as access to a clean source of drinking water or sanitation). Given data limitations, this can substantially reduce the set of possible indicators.

In this case, group-specific measures aim to capture achievement or deprivations of relevant groups (eg. children, women or people with disabilities). This approach has important policy advantages, as these measures can provide accurate information specific to these groups and their needs.However, if the applicable population is a subgroup, it can no longer serve as a national measure to track multidimensional poverty. In addition, group-specific measures might miss possible overlaps between disadvantaged groups (e.g. women, people with disabilities and minority ethnic groups).

Step 4: Choice of dimensions and indicators

Include universally applicable achievements

Test assumptions about household distribution

Combined measures use achievements from a subset of household members, making assumptions about how achievements are distributed among household members. These assumptions should be clearly stated and justified by available evidence and theory. Most countries have used this option, combined measures, to design their national MPls. In doing so, they have rigorously addressed two practical challenges:

1. Some households do not have members within the applicable population for certain indicators. For example, households with no small children do not have any members with information on vaccinations; similarly, households with no school-aged children have no members with information on school attendance.
2. Data is missing or not collected for some household members, even though they are part of the applicable population. For instance, though nutrition is a relevant issue for every individual, household surveys may only cover children under five and women of reproductive age. Household members outside of these groups may not be measured for anthropometrics, and households with no small children or women in reproductive age will have no information on nutrition at all.

As datasets are improving, these limitations are falling sharply and measures are becoming more accurate. In the meantime, special studies of omitted populations (such as children, the elderly or indigenous peoples) can complement poverty figures obtained from the national MPI.

In this case, the measure would only include achievements applicable to the whole population (such as access to a clean source of drinking water or sanitation). Given data limitations, this can substantially reduce the set of possible indicators.

Step 4: Choice of dimensions and indicators

Income as an indicator in a national MPI

During the process of designing a national MPI, different aspects should be discussed when considering whether or not to include income as an indicator. Hover over the boxes below to learn more.

The purpose of the measure

Data sources for the measures

Matches and mismatches

Measurement error

Sources of data available

Step 1: Purpose

Step 7:Poverty cut-off

Step 2:Space

Step 8:Compute

Step 5: Deprivations

Step 4: Dimensions

Overview

Step 3: Identification

AFMethod

Course structure

Welcome

Lesson 3

Lesson 3 completed

Resources/References

Step 6: Weights

Step 5: Setting deprivation cut-offs

Click on the icons below to learn about the steps for setting deprivation cut-offs.

step 2

step 3

step 1

Compute the uncensored headcount ratios

Run a preliminary analysis

Create a large set of potential indicators

Click the arrows to learn more about Step 5.

Step 1: Purpose

Step 7:Poverty cut-off

Step 2:Space

Step 8:Compute

Step 5: Deprivations

Step 4: Dimensions

Overview

Step 3: Identification

AFMethod

Course structure

Welcome

Lesson 3

Lesson 3 completed

Resources/References

Step 6: Weights

Step 5: Setting deprivation cut-offs

Create a large set of potential indicators

The definition of the deprivation cut-off for each indicator is a normative decision. However, the technical team needs to check how the available data reflect normative decisions. It may be helpful to start by creating a large set of potential indicators, some of them closely related to each other but using different deprivation cut-offs. This is necessary to assess the sensitivity of measures to a change in deprivation cut-offs and also, in the case of uncertainty about which cut-off to choose, to clarify the implications of a choice to policy users.

Step 5: Setting deprivation cut-offs

Compute the uncensored headcount ratios

Once the list of potential indicators has been created, it is possible to compute the uncensored headcount ratios (i.e., the proportion of people deprived in each indicator) for all indicators and to compare the different versions of the same indicator with different deprivation cut-offs. Indicators with slightly different deprivation cut-offs should have similar uncensored headcount ratios. For example, a household is deprived if no member has at least 6 years of schooling or a household is deprived if at least one member does not have 6 years of schooling.

Step 5: Setting deprivation cut-offs

Run a preliminary analysis

Before finalizing indicators into the national MPI, it is useful to run a preliminary analysis and understand the relationship between indicators. This might lead to dropping an indicator, to combining some indicators into a sub-index or to adjusting the categorization of indicators into dimensions.

Step 5: Setting deprivation cut-offs

Video: Setting deprivation cut-offs in Bhutan Example

Watch the video to hear Sabina Alkire describe a field study in Bhutan designed to collect input into a draft of the national MPI, including the deprivation cut-offs.

Step 1: Purpose

Step 7:Poverty cut-off

Step 2:Space

Step 8:Compute

Step 5: Deprivations

Step 4: Dimensions

Overview

Step 3: Identification

AFMethod

Course structure

Welcome

Lesson 3

Lesson 3 completed

Resources/References

Step 6: Weights

Step 5: Setting deprivation cut-offs

Weights, like dimensions, indicators, and cut-offs, are fixed over time. Setting weights plays a fundamental role in defining the relative importance of each deprivation in the final measure. Click on each segment of the graph learn more about the different arguments for setting weights.

Empirical

Arguments for setting weights

Normative

Household

Step 1: Purpose

Step 7:Poverty cut-off

Step 2:Space

Step 8:Compute

Step 5: Deprivations

Step 4: Dimensions

Overview

Step 3: Identification

AFMethod

Course structure

Welcome

Lesson 3

Lesson 3 completed

Resources/References

Step 6: Weights

Step 5: Setting deprivation cut-offs

Arguments for setting weights

The selection of the weighting structure of a measure can be justified using normative arguments reflecting the context and priorities of the country, but empirical tests of robustness are also essential. Both types of arguments work together as neither is complete alone. Depending on the purpose of the measure, different alternatives of weights should be considered:

• If a measure aims to provide evidence for policy, it is important that different stakeholders understand all parameters of the measure, and are able to communicate those to civil society. In this case, normative weights may be the best option.
• If a measure is an academic exercise, the use of empirical weights can be recommended because over time comparability or communication are not a priority.

The selection of weights has important implications on the contribution of each indicator to the MPI and how the measure changes over time. Although the use of these methods provides important information regarding the rankings that poor individuals give to their own deprivations, they can also be affected by personal preferences and situations. No country has yet used the participatory approach to assign weights to different indicators. However, in Bhutan, participatory exercises were conducted after the design of the measure to validate the rank of deprivations used in the national MPI.

Step 5: Setting deprivation cut-offs

Empirical

Empirical arguments are associated with different statistical techniques, including regression analysis, frequency-based weighting and multivariate statistical weighting. Different aspects should be considered when using this method:

• Although these weights are considered as "objective" because they are the results of the data, the final results of the MPI are difficult to communicate to policymakers. For example, the final weighing structure of the index varies depending on the sample used and the method used to calculate the weights, which makes communication of the results of the MPI difficult.
• Statistical weights depend on the data structure. Because they analyze correlations between different indicators, which depends on the information collected in the survey, statistical weights change between surveys. Therefore, the structure of the MPI will vary with every wave of the data and it will not be possible to track progress over time.

Step 5: Setting deprivation cut-offs

Normative

Normative arguments are related to the relative importance that policymakers or other stakeholders place on each indicator in the measure. Given that an MPI is a social evaluation, there may need to be consensus on a range of weights. In the case of normative arguments, the justifications are usually easy to understand and communicate to the public, and, once decisions have been made, the structure of the index remains stable over time, which brings credibility and sustainability.

Step 6: Data sources for setting weights

Video: Data sources for setting the weights of an MPI

Two popular sources of data for setting the weights are:

• Participatory exercises (focus groups)
• Surveys

Watch the video to hear Sabina Alkire elaborate on each of these options.

Step 1: Purpose

Step 7:Poverty cut-off

Step 2:Space

Step 8:Compute

Step 5: Deprivations

Step 4: Dimensions

Overview

Step 3: Identification

AFMethod

Course structure

Welcome

Lesson 3

Lesson 3 completed

Resources/References

Step 6: Weights

Step 7: Approaches to defining the poverty cut-off

How much is enough to be considered poor?

The poverty cut-off k reflects the minimum level of deprivations or deprivation score an individual or household must be suffering simultaneously to be considered poor.The setting of this poverty line needs to reflect the priorities and policy goals of the country. For example, if the purpose is to target resources to the poorest populations, the value of k must capture those who are facing the highest number of simultaneous deprivations. If the goal is to monitor poverty, k should reflect the minimum level of deprivations acceptable in a country to be considered non-poor. Different approaches have been presented in the literature and in practice to define the poverty cut-off. Click on the boxes below to learn more about each approach.

Intersection approach

Union approach

Dual cut-off approach

Step 1: Purpose

Step 7:Poverty cut-off

Step 2:Space

Step 8:Compute

Step 5: Deprivations

Step 4: Dimensions

Overview

Step 3: Identification

AFMethod

Course structure

Welcome

Lesson 3

Lesson 3 completed

Resources/References

Step 6: Weights

Step 7: Approaches to defining the poverty cut-off

Union approach

The union approach identifies people as multidimensionally poor if they experience at least one measured deprivation. This approach is an important tool when the purpose of the measure is related to advocacy, as it clearly identifies people facing any given deprivation and increases the visibility of poverty, because it identifies a larger number of people as poor. The risks of using the union approach involve:

• If the purpose is related to policy, this approach might overestimate the number of people or households who are multidimensionally poor, as a single observed deprivation might be in fact representing individual preferences. Also, as the number of total indicators included in the MPI increases, the likelihood of being deprived in just one of them also rises, increasing the incidence of poverty as the national MPI.
• Given that the union approach only uses one indicator to define who is or not poor, the multidimensionality of poverty can be lost.

Step 7: Approaches to defining the poverty cut-off

Intersection approach

The intersection approach identifies as poor only those individuals who are deprived in all the indicators simultaneously. This leads to low incidence rates as it only captures individuals who are extremely poor. In particular, as the number of indicators increases, the likelihood of being deprived in all of them at the same time reduces.

Step 7: Approaches to defining the poverty cut-off

Dual cut-off approach

The Alkire Foster method uses a dual cut-off approach, which sets a poverty line k that can range between 1 and the total number of indicators included in the measure. The dual cut-off includes the union and intersection approaches as specific cases, as well as intermediate poverty cut-offs.

Summary

There is no universal rule for defining the poverty cut-off. Ordinarily, the poverty cut-off reflects the weighting structure of the index.For example, if there are three dimensions, it would be natural to explore cut-offs of 33 percent and 34 percent—which translate to being deprived in at least one dimension or deprived in more than one dimension or the equivalent of weighted indicators. In any case, the selection of the poverty cut-off is a normative decision that needs to be easy to communicate and statistically tested.

Step 8: Compute the MPI

Video: An intuitive explanation of the Alkire Foster method

Watch the video to hear Christian Oldiges, Co-Director of Metrics and Policy at OPHI, share a simple explanation of the Alkire Foster method.

Click the arrows to learn more about Step 8.

Step 1: Purpose

Step 7:Poverty cut-off

Step 2:Space

Step 8:Compute

Step 5: Deprivations

Step 4: Dimensions

Overview

Step 3: Identification

AFMethod

Course structure

Welcome

Lesson 3

Lesson 3 completed

Resources/References

Step 6: Weights

Step 8: Compute the MPI

Click on each box below to learn about the different steps of the Alkire Foster method.

Compute MPI

Apply weights

Select structure

Step 3

Step 1

Step 2

Step 4

Step 5

Create deprivation profile

Aggregate information

Step 1: Purpose

Step 7:Poverty cut-off

Step 2:Space

Step 8:Compute

Step 5: Deprivations

Step 4: Dimensions

Overview

Step 3: Identification

AFMethod

Course structure

Welcome

Lesson 3

Lesson 3 completed

Resources/References

Step 6: Weights

Step 8: Compute the MPI

Select structure

Start by selecting the structure of the measure:

• Purpose of the index
• Space of the measure
• Unit of identification and analysis (individual/household)

Step 8: Compute the MPI

Create deprivation profile

For each of the indicators included in the MPI, compare the achievement of the individual to the respective deprivation cut-off and classify the individual as deprived or non-deprived. For example, start by examining whether the individual has adequate sanitation, access to drinking water, access to health services, if all children attend school, if adults have decent work, etc. The set of indicators is flexible and can be adapted to the context in which poverty is being measured. In summary, take the following steps:

• Set the deprivation cut-offs for each indicator
• Compare the individuals’ achievements with the indicators’ deprivation cut-offs: after applying these cut-offs, each individual is identified as either deprived or non-deprived in each indicator.

Step 8: Compute the MPI

Apply weights

Apply weights (which must add up to one, or 100 percent) to each of the deprivations, and then sum them so that each person or household has a deprivation score that gives the weighted percentage of deprivations they experience. Next, identify people as multidimensionally poor if the weighted sum of their deprivations is greater than or equal to the poverty cut-off—which might be, for example, 20 percent, 33 percent or 50 percent.

Step 8: Compute the MPI

Apply weights

In summary, take the following steps:

1. Set the value of the weights per each dimension and indicator
2. Apply these weights to the deprivation matrix to obtain the weighted deprivation matrix (multiply each deprivation by the weight of that indicator)
3. Set the poverty cut-off (known as k) to identify who is poor. There is no one way to set k, considering the results with all possible values of k to determine whether results are robust to the choice of k. As we will see later in the course, the selection of the poverty cut-off will depend on aspects related to the purpose of the measure, communication of the results among other reasons.
4. Create the censored deprivation matrix which focuses only on the deprivations of the poor. To do this, you 'censor' or replace with zero the entries for all indicators of all people who have not been identified as poor (that is, those whose deprivation score is below the poverty cut-off k). The censored deprivation matrix is used to make the MPI and its associated information. For example, the censored headcount ratio for an indicator is the mean of that indicator's column in the censored deprivation matrix.

Step 8: Compute the MPI

Aggregate information

Aggregate information into two informative indices:

1. The incidence of poverty (H), which is the proportion of people identified as multidimensionally poor, also referred to as the "headcount ratio". It is the percentage of people out of the total population whose weighted deprivation score is greater than or equal to the poverty cut-off.
2. The intensity of poverty (A), which is the average proportion of indicators in which poor people are deprived—the average deprivation score across all poor people.

At this stage, take the following steps:

1. To compute the incidence of poverty, divide the number of people identified as multidimensionally poor by the total number of people.
2. To compute the intensity of poverty, calculate the average deprivation share among the people who were identified as poor: add up the deprivation scores of the poor and divide them by the total number of poor people.

Step 8: Compute the MPI

Compute the MPI

The MPI (also referred to as M0) is computed as the product of two components - the incidence and intensity of poverty [MPI = H x A]. Usually H and A are written as percentages, but the MPI is written as an index, usually with three digits. This method not only identifies who is poor but also innovates by incorporating how acute or intense the situation of multidimensional poverty is for the poor.

The MPI can be read as the percentage of deprivations that poor people experience out of the total deprivations that would be experienced by the society if all people were deprived in all indicators simultaneously. At this stage, take the following step: 1. Multiply the incidence of multidimensional poverty by the intensity of multidimensional poverty. H x A. (Please refer to this week's lecture for a detailed explanation of the Alkire—Foster (AF) method.)

Alkire-Foster Method

Example of the Alkire-Foster method

Suppose you are interested in analyzing the multidimensional poverty of a hypothetical society with four people along four indicators:

1. Hectares of land
2. Years of schooling
3. Body mass index (BMI)
4. Access to improved sanitation

Go over each step of the MPI computation for this example on the next slide. Since the structure of the measure is provided to us in this case, start with Step 2 of the Alkire-Foster method: creating a deprivation profile.

Step 1: Purpose

Step 7:Poverty cut-off

Step 2:Space

Step 8:Compute

Step 5: Deprivations

Step 4: Dimensions

Overview

Step 3: Identification

AFMethod

Course structure

Welcome

Lesson 3

Lesson 3 completed

Resources/References

Step 6: Weights

Alkire-Foster Method

Example of the Alkire-Foster method

Starting at Step 2, click on each box below to learn about the different steps of the Alkire Foster method in this example.

Compute MPI

Apply weights

Step 3

Step 2

Step 4

Step 5

Create deprivation profile

Aggregate information

Step 1: Purpose

Step 7:Poverty cut-off

Step 2:Space

Step 8:Compute

Step 5: Deprivations

Step 4: Dimensions

Overview

Step 3: Identification

AFMethod

Course structure

Welcome

Lesson 3

Lesson 3 completed

Resources/References

Step 6: Weights

Alkire-Foster Method

Example: Create a deprivation profile

Note the deprivation cut-off: The deprivation cut-off vector is denoted by z = (5, 5, 18.5, has access to improved sanitation) and is used to identify who is deprived in each indicator. For instance, a person who has not completed five or more years of schooling is considered deprived in education. Similarly, a person is deprived in sanitation if she does not have access to improved sanitation in her home.

Alkire-Foster Method

Example: Create a deprivation profile

Compare the individuals' achievements with the indicators’ deprivation cut-offs: By doing so, we construct the deprivation matrix 90, where a cell has the score of 1 if the person (row) is deprived in the indicator (column), and a score of 0. For ease of interpretation, all achievements that are below the corresponding deprivation cut-off in the matrix above are underlined; those entries are now replaced by 1 (i.e. deprived) in the 90 matrix.

Alkire-Foster Method

Example: Apply weights

Note the value of the weights: All indicators are equally weighted at one quarter, and thus the weight vector is w = (0.25, 0.25, 0.25, 0.25).

Alkire-Foster Method

Example: Apply weights

Apply these weights to the deprivation matrix to obtain the weighted deprivation matrix (90). The weighted sum of the deprivations is the deprivation score (ci) of each person. For example, person 1 has no deprivations and so the deprivation score is 0, whereas person 3 is deprived in all indicators and thus has the highest deprivation score of 1. Similarly, the deprivation score of the second and fourth people are 0.5 (0.25 + 0.25) and 0.25, respectively.

Alkire-Foster Method

Example: Apply weights

Identify the poor, set the poverty cut-off: A poverty cut-off, denoted by k, is used to identify who is poor. For example, if k is 0.50 or 50 percent, then a person is poor if he/she is deprived in half or more of the weighted indicators (if his/her deprivation score is 0.50 or higher). In this case that would mean that two of the four people are identified as poor (i.e. persons 2 and 3). The poverty cut-off could be set to include two special cases: union and intersection identification, but in practice, an intermediate value usually is used. All national MPIs to date have used an intermediate value for the poverty cut-off rather than an intersection or union approach.

Alkire-Foster Method

Example: Apply weights

Calculate the censored deprivation matrix to focus only on the deprivations of the poor: once the poor have been identified in Step 3, deprivations of people who were identified as non-poor are replaced with a zero. This leads to the censored deprivation matrix (g0(k)) and the censored deprivation score in which every deprivation belonging to a non-poor person is set to zero.

In our example, we chose a poverty cut-off k of 0.50. There is one case in which censoring is not relevant: when the poverty cut—off corresponds to the union approach. In this case, any person deprived in any indicator is considered poor. Therefore, no censoring is needed and both the censored and the original matrix are identical.

Alkire-Foster Method

Example: Aggregate the information

Aggregate the incidence of poverty: the headcount ratio H is the proportion of people who are poor, which is two out of four people in our matrix. That is, H = 2/4 = 1/2. This means that 50 percent of the people in this example are multidimensionally poor.

Aggregate the intensity of poverty: the intensity A is the average deprivation share among the poor, which in this example is the average of 0.5 and 1 (i.e. the deprivation scores of the two people who are poor, persons 2 and 3). That is, A = 75 percent. Thus, multidimensionally poor individuals are deprived in 75 percent of weighted indicators, on average.

Alkire-Foster Method

Example: Compute the MPI

The MPI can then be obtained as the product of the incidence and the intensity of poverty. That is, MPI = H x A = 50% x 75% = 0.375. In this example, the MPI means that multidimensionally poor people experience 37.5 percent of the total deprivations that would be experienced if all people were deprived in all indicators at the same time.

Alkire-Foster Method

Important properties of the Alkire Foster measures

Population subgroup decomposability

Dimensional breakdown

The MPI is made up of many indicators and it can be unfolded very precisely, by exactly these same indicators. More specifically, MPI is H x A, but another way you can compute MPI is to take the share of people who are poor and are deprived in each indicator (e.g., 20 percent in health, 30 percent in education, 10 percent in work), multiply these by their weights (eg. one third each), and the "answer" is the MPI (in this example, MPI = 0.200).This property allows the composition of multidimensional poverty to be analyzed.

Population subgroup decomposability means that the MPI, H, A and each indicator can be disaggregated by any group for which the data are representative. This is a key priority for the SDGs.In practice, if the MPI is disaggregated to show subgroup poverty levels, then these add up, using population shares, to the national figures. This feature has proven to be of great use in analyzing poverty by regions, by gender, by age and by other relevant subgroups.

Step 1: Purpose

Step 7:Poverty cut-off

Step 2:Space

Step 8:Compute

Step 5: Deprivations

Step 4: Dimensions

Overview

Step 3: Identification

AFMethod

Course structure

Welcome

Lesson 3

Lesson 3 completed

Resources/References

Step 6: Weights

Lesson 3 completed

Congratulations on completing Lesson 3! This lesson described the steps involved in designing a national MPI, from the definition of the unit of analysis to the selection of the poverty cut-off. It also introduced you to the Alkire-Foster method of multidimensional poverty measurement and the steps involved in computing an MPI using this method.Lesson 4 will further explore the technical process of designing a national MPI. Before you proceed to Lesson 4, we invite you to test your knowledge of the information covered in Lesson 3 by taking Quiz 3.

Step 1: Purpose

Step 7:Poverty cut-off

Step 2:Space

Step 8:Compute

Step 5: Deprivations

Step 4: Dimensions

Overview

Step 3: Identification

AFMethod

Course structure

Welcome

Lesson 3

Lesson 3 completed

Resources/References

Step 6: Weights

Additional resources

Lecture by Sabina Alkire "Normative Issues in Multidimensional Poverty Measurement" (Video)

"Multidimensional Poverty Measurement & Analysis" (book, PDF)

Seminar recording "Case studies: Multidimensional Poverty Index" (video)

Lecture by Sabina Alkire "Identification and Aggregation in the Alkire Foster Method" (video)

Lecture by James Foster "The Alkire Foster Methodology" (video)

Step 1: Purpose

Step 7:Poverty cut-off

Step 2:Space

Step 8:Compute

Step 5: Deprivations

Step 4: Dimensions

Overview

Step 3: Identification

AFMethod

Course structure

Welcome

Lesson 3

Lesson 3 completed

Resources/References

Step 6: Weights

References

Lecture by Sabina Alkire “Identification and Aggregation in the Alkire Foster Method"

Handbook "How to build a National Multidimensional Poverty Index (MPl)”

Atkinson, Anthony & Marlier, Eric. (2010). "Analysing and Measuring Social Inclusion in a Global Context”

Lecture by Sabina Alkire “Normative Issues in Multidimensional Poverty Measurement"

Step 1: Purpose

Step 7:Poverty cut-off

Step 2:Space

Step 8:Compute

Step 5: Deprivations

Step 4: Dimensions

Overview

Step 3: Identification

AFMethod

Course structure

Welcome

Lesson 3

Lesson 3 completed

Resources/References

Step 6: Weights

References

Decancq, K. and Lugo, M. A. (2008). "Setting Weights in Multidimensional Indices of Well—Being." OPHI Working Paper 18, University of Oxford.

Burki, Abid. (2011). Exploring the Links between Inequality, Polarization and Poverty: Empirical Evidence from Pakistan.

Sarmento, Rui. (2020). " Regression Model for GDP Growth from UN (United Nations 2018 DATASET)"

Step 1: Purpose

Step 7:Poverty cut-off

Step 2:Space

Step 8:Compute

Step 5: Deprivations

Step 4: Dimensions

Overview

Step 3: Identification

AFMethod

Course structure

Welcome

Lesson 3

Lesson 3 completed

Resources/References

Step 6: Weights