Colombia’s Multidimensional Poverty Index
Colombia – a pioneering new poverty plan
In 2011, the Government of Colombia adopted a pioneering new poverty-reduction strategy, which sets firm and binding targets to close the country’s poverty gaps. Devised by Colombia’s Ministry of Planning, it is the first national poverty reduction plan to use the Alkire Foster (AF) method for measuring multidimensional poverty, developed by OPHI.
Underlying these targets is a new national Multidimensional Poverty Index for Colombia (the MPI-Colombia). This new poverty measure uses an innovative adaptation of the AF method, customising the dimensions and indicators to the country’s specific needs and public policy priorities. This multidimensional poverty measure underpins the country’s ambitious poverty reduction goals. It is a powerful example of how this methodology can inform poverty reduction strategies and help to create a clear system of accountability.
Building on the flexibility inherent in the AF method, the MPI-Colombia assesses broader social and health-related aspects of poverty in five dimensions:
- Household education conditions
- Childhood and youth conditions
- Access to household utilities and living conditions
The five dimensions are equally weighted and use 15 indicators.
The multidimensional poverty measure developed by the Colombian Government forms part of a comprehensive poverty reduction strategy. The government plans to reduce multidimensional poverty by 13 per cent– from 35 per cent of the entire population in 2008 to 22 per cent in 2014.
The measure is being used twice: once to set the targets and second to track progress towards them. This approach powerfully shows the way the method can be used to help inform poverty reduction strategies. The case of Colombia shows that in practice, the measure is a powerful monitoring and evaluation tool, as well as a flexible measure of poverty or wellbeing.
To ensure that the targets are on track, the Government is using a “traffic light” system which triggers alerts when progress towards each indicator falls off track. The targets are drawn from explicit goals set out in the 2011 National Development Plan. The household surveys used to calculate MPI-Colombia have historical information from 1997, 2003 and 2007. Since 2010 the survey has been collected annually.
Heart of Government
The new National Development Plan was introduced with support at the highest levels of the Government. Colombian President, Juan Manuel Santos, used the launch of the new poverty plan as an opportunity to affirm his personal commitment to making poverty reduction the centrepiece of his government – and to building on the country’s substantial progress in social reform in the last decade. The country’s wider development plan has three pillars: employment, poverty reduction and security. President Santos has made poverty reduction the top priority out of these three.
In being embraced at the heart of the Government, the MPI-Colombia shows how a poverty measurement methodology can be used to create a clear accountability system. In Colombia, this involves the ministries of Education, Social Protection, Housing, and the other main actors involved in social policy. Hernando José Gómez, Director of the National Planning Department (DNP), has said of the measure:
“The Government of Colombia hasn’t just created a new poverty measure – we have created a comprehensive new poverty strategy. This has been embraced into the heart of the government – as concrete goals, as a mechanism of accountability, and as a robust test of our extreme poverty reduction strategy.”
Three-pronged poverty reduction plan
The MPI-Colombia is part of a wider group of tools which underpin Colombia’s 2011 poverty strategy, including a new income poverty line and a new Commission on Social Mobility.
The table below shows Colombia’s multidimensional poverty reduction plan in detail – both the dimensions and indicators of the MPI-Colombia and the National Development Plan baseline for 2008 and the goal (by 2014).
The Alkire-Foster method
This technique was developed by Sabina Alkire, OPHI Director, and James Foster, OPHI Research Associate and Professor of Economics and International Affairs at George Washington University. The Alkire-Foster method measures outcomes at the individual level (person or household) against multiple criteria (dimensions and indicators). The method is flexible and can be used with different dimensions, indicators, weights and cutoffs to create measures specific to different societies and situations. Read more about the method and how to calculate it: Alkire-Foster method.
Inside Colombia’s multidimensional poverty reduction strategy
Dimensions, indicators and targets (weights are given in brackets)
(weight in brackets)
(weight in brackets)
|Indicator National Development Plan||
Baseline NDP 2008
Data for 2010
|Goal NDP 2014 (%)|
|Education conditions (for households) (0.2)||Educational achievement (0.1)||Average education level for people 15 and older living in a household.||Low educational achievement at the household level||58.8||55.4||52.8|
|Literacy (0.1)||Percentage of people living in a household 15 and older who can read and write.||Illiteracy rate for population 15 and older||14.2||13.2||12.0|
|Childhood and youth conditions (0.2)||School attendance (0.05)||Percentage of children between the ages of 6 and 16 that attend school.||Non-assistance rate for population from 6 to 16.||5.4||4.6||3.5|
|No ‘school lag’ (children older than the average age in a given school year) (0.05)||Percentage of children and youths (7-17 years old) within the household not subject to school lag (according to the national norm)||School lag for population from 7 to 17.||33.4||35.1||33.1|
|Access to child care services (0.05)||Percentage of children between the ages of 0 and 5 who simultaneously have access to health, nutrition and education.||Barrier to access of child care services||11.2||10.8||9.2|
|Children not working(0.05)||Percentage of children not working (ie. subject to child labour).||Child work for children from 12 to 17 years old||8.2||6.8||5.6|
|Employment (0.2)||No one in long- term unemployment (0.1)||Percentage of household members from the economically active population (EAP) who don’t face long-term unemployment (more than 12 months).||Long term unemployment rate||9.6||9.9||9.3|
|Formal employment (0.1)||Percentage of household ́members from the economically active population (EAP) employed and affiliated to a pension fund (this indicator is used as a proxy for whether people are formally or informally employed)||Informality rate||80.6||80.9||74.7|
|Health (0.2)||Health insurance (0.1)||Percentage of household members over the age of 5 that are insured by the Social Security Health System||No health insurance||24.2||21.0||0.5|
|Access to health services (0.1)||Percentage of people within the household that have access to a health institution in case of need||Access barriers to health services||8.9||6.9||2.4|
|Access to public utilities and housing conditions (0.2)||Access to water source (0.04)||Urban household: considered deprived if lacking public water system. Rural household: considered deprived when the water used for the preparation of food is obtained from wells, rainwater, spring source, water tank, water carrier or other sources.||Low coverage of pipe water||12.9||11.6||10.9|
|Adequate elimination of sewer waste (0.04)||Urban household: considered deprived if lacking public sewer system. Rural household: considered deprived if uses a toilet without a sewer connection, a latrine or simply does not have a sewage system.||Low coverage of sewer waste||14.1||12.0||11.3|
|Adequate floors (0.04)||Lacking materials (dirt floors)||Inadequate floors||7.5||6.3||5.6|
|Adequate external walls (0.04)||An urban household is considered deprived when the exterior walls are built of untreated wood, boards, planks, guadua or other vegetable, zinc, cloth, cardboard, waste material or when no exterior walls exist. A rural household is considered deprived when exterior walls are built of guadua or another vegetable, zinc, cloth, cardboard, waste materials or if no exterior walls exist.||Inadequate walls||3.1||3.0||2.1|
This section draws substantially on Angulo, Pardo and Diaz (2010).
Unit of analysis
The household was selected as the unit of analysis for the MPI-Colombia. Household members are therefore considered to be deprived according to the achievements of all household members simultaneously. For example, a person is considered to be deprived if any of their household members is deprived in literacy.
To select the unit of analysis, three criteria were used.
The first, a normative criterion draws on the Colombian political constitution, which claims that the guarantee of living conditions and rights is the joint responsibility of the family, society and the State – not the responsibility of individuals in isolation.
The second empirical -criteria draws on academic evidence relating to Colombia which shows that households historically respond to adverse situations collectively. For example, the Misión Social (Social Mission, 2002) found that during the crisis in the 1990s in Colombia, unemployment of the household head had the single biggest impact on household, while the main recovery strategy was the entry of the spouse and children into the labour market.
The final criteria, meanwhile, relates to the social policy context of the country. This criterion draws on existing policies, programmes and instruments in the country – such as SISBEN, UNIDOS network strategy, Familias en Accion (a conditional cash transfer programme) which all are all focused at the household level.
Criteria for the selection and validation of variables
The variables of the MPI-Colombia were selected using the following three criteria: frequency of use, ability to show alterations though public policy decisions, and availability within the Colombian Living Standards Measurement Survey (LSMS).
The frequent use criterion drew on a literature review of several multidimensional measurements of wellbeing or deprivations (national and international) and consultations with experts. The consultation with experts ensured that the variables selected were regularly used by ministries and agencies to track overall public policies. The LSMS was selected as the source of information because of the richness of information included and the frequency of its publication.
The variables that met the above three criteria were then validated as appropriate if they did not have a relative error bigger than 15%.
The MPI-Colombia uses a nested weighting structure where each dimension has the same weight (0.2) and each variable has the same weight within each dimension. This set of weights was selected to reflect the equal importance of each dimension as a constituent element of quality of life. Other sets of weights were developed by the Colombian team but they did not receive consensus in debates with experts. Moreover, the consultation also showed that the debate on the interaction, complementarity and substitutability across variables is underdeveloped at this time.
The overall poverty threshold – ‘k’ – was set at one-third of the weighted dimensions. This parameter – k – gives the share of dimensions in which a person must be deprived in order to be considered multidimensionally poor. In order to select the value for k, Colombia used statistical criteria combined with analytical validation. Income poor persons on average experienced just over one-third of the deprivations; similarly the people who perceived themselves to be poor experienced one-third of the deprivations. The people who were income poor and perceived themselves to be poor experienced a slightly higher share of deprivations. The statistical analysis included computing poverty for all k values and systematically checking the robustness of the results to changes in the k value.
First of all, a robust band of k values was devised; this robust band relied on the precision of the sample, avoiding relative errors bigger than 15% for the poverty measures (H, M0, M1). Next, dominance exercises were performed to check the robustness of the results. Furthermore, statistical mean tests were used to avoid the overlapping of confidence intervals for the poverty measures using different k values.
Finally, in order to reach consensus on the number of deprivations a person must face to be considered multidimensionally poor, the median and the average number of deprivations across several indicators where checked (see the table below). The parameter k = 5 was agreed upon as a result.
H = The percentage of the population who are poor – or the (raw) Headcount ratio.
A= The average intensity of deprivation.
M0= An ‘adjusted Headcount’. This reflects both the incidence (the percentage of the population who are poor, H) and intensity of poverty, A.
M1= This measure reflects the incidence, intensity and depth of poverty. The depth of poverty is the ‘gap’ (G) between poverty and the poverty line (M1 = H x A x G).
Number of deprivations experienced, though different groups of population, 2008
|Median of deprivations||Average number of deprivations|
|Population which perceives itself as poor||5.0||5.0|
|Population below the income poverty line||5.1||5.2|
|Population which perceives itself as poor and is below the income poverty line||5.4||5.6|
|Non-poor population by perception||3.0||3.2|
|Population over income poverty line||3.0||3.2|
Source: NPD-SDQLD calculations, with data from the LSMS 2008
APPLICATIONS ~The MPI-Colombia as public policy design tool
MPI-Colombia at the municipal level
A proxy of the national MPI-Colombia was constructed at the municipal level using Census data from 2005. The municipal MPI allowed poverty maps to be created and updated using the new multidimensional approach and assessment tool. The exercise is a rich source of information for targeting purposes at the geographical level.
MPI-Colombia goals for national planning
The MPI-Colombia has three key policy advantages: it can depict the joint distribution of deprivations (the intensity of poverty), it shows the contribution of each of the dimensions to overall poverty levels, and is able to facilitate accountability at the level of social public policy interventions.
On the strength of these advantages, the Colombian government decided to include the multidimensional (raw) Headcount ratio (H) within the indicators used to track poverty reduction strategies in the National Development Plan –NDP (2010-2014).
The NDP set a goal to reduce the population living in multidimensional poverty from 35% (2008) to 22.5% (2014). The overall 2014 poverty reduction goal was established according to goals attached to each of the MPI dimensions. Consequently, the overall target is a combination of the achievements of each of the five dimension-specific strategies designed in the NDP. This institutional arrangement makes the MPI-Colombia a powerful monitoring tool and a comprehensive poverty reduction strategy.
Targeting beneficiary households for the Extreme Poverty Reduction Strategy – UNIDOS
In Colombia the main public policy initiative to reduce extreme poverty is the UNIDOS network. This network combines the efforts of several governmental agencies and affected families with the goal of enhancing the income-generating abilities and the quality of life conditions of the families. This intervention is by nature transitory; once a family no longer lives in extreme poverty, the family finishes the programme. The MPI-Colombia is now used alongside income poverty measures as a condition for promoting families from the UNIDOS programme. That means families classified as non-multidimensionally poor (using the H headcount ratio with k=5) and non-income poor, will be promoted out of the programme at the end of the process.
 The selection of the indicators considered the previous Colombian multidimensional experiences as the Unmeet Basic Needs, the SISBEN III and the Quality of Life Index.