Do we need the Multidimensional Poverty Index (MPI) for post-2015?



It was very exciting to witness the shift in the understanding of poverty of many governments, who have adopted a more multidimensional measurement of poverty. The Oxford Poverty and Human Development Initiative (OPHI), led by Sabina Akire, can claim the great achievement of having pushed over 20 countries to either adopt or experiment with new ways of measuring poverty. The OPHI played a key role in changing the conceptualisation of poverty amongst policymakers.

Inaugurated in the 2010 Human Development Report, the MPI has contributed to opening the debate on the necessity of developing indicators that could better capture different deprivations. The MPI measures three dimensions of poverty – health, education and living standards – through 10 indicators. It is calculated by multiplying the number of people identified as multidimensional poor by the average percentage of deprivations that poor people experience. In this way, it takes into account both the proportion of poor people and the intensity of their poverty.

During a side-event at the last UN General Assembly on post-2015, a new global MPI 2.0 was proposed as the headline indicator for the new development framework. Ministers from important countries clearly acknowledged the limitations and insufficiency of income approaches and recognised the importance of having multiple indicators to capture different dimensions of poverty. It was a belated but very welcome step forward.

The Director of the Mexican National Council for the Evaluation of Social Development Policy, showed how by looking at all the indicators for specific regions, Mexican policymakers can assess the impact of public policies. Moreover, they can also identify the priority sectors for interventions in different regions of the country.

The Statistician General at the Nigerian National Bureau of Statistics went a step further, arguing how useful it was for them to be able to include a subjective indicator of poverty. This revealed that, in Nigeria, many more people than those captured by official statistics would consider themselves as poor. People with valuable assets such as land and cattle could still live in poor housing conditions without basic services, issues that need to be addressed by public policies.

All the presentations by the different governments highlighted the importance of having data for these indicators in different areas, but not necessarily of having them pulled together into a single index. Do we really need to aggregate them into a single global indicator? What is the added value of this operation?

I was often told that the power of the poverty headcount ratio (often called the dollar a day poverty) lies in its simplicity and the fact that it is just one number. Is the construction of such a complex indicator going to help in the eradication of poverty?

Using non-monetary deprivation indicators to analyse poverty and social exclusion in Europe has shown how slight changes in the items selected can lead to huge differences in the results (click here for an example). To gain insights useful for national planning, indicators utilised for the measurement of poverty needs a lot of adaptation in just one country and often use different methodologies. Another challenge is to identify universal indicators for issues that might not imply similar levels of deprivation in different contexts. For instance, not having a floor in the house is not an indicator of deprivation for a pastoralist community.

There is also a problem around establishing cut-offs and the weight of each indicator. What is the cut-off at which we consider someone deprived or not? How many deprivations are needed to consider someone to be multidimensional poor? How can we establish that a slightly higher number of people who have a floor in their shelter correspond to a certain number of years of education? These are all arbitrary and problematic decisions which make the MPI indicator a confused aggregation of apples and oranges.

Duncan Green argues that we need league tables to set governments competing against each other in order to achieve development. Amartya Sen’s critical assessment of India’s development progress, while praising the significant achievements of Bangladesh, despite its lower income, may effectively push governments to act. But how would you act to increase your score under an MPI index?

A government may work strategically on the cheaper deprivations to move the number of people needed beyond the cut-off for specific indicators. Moreover, a large component of the Global MPI is based on input indicators regarding education, such as, school attendance and years of schooling which simply measure that fact that a number of pupils are put in a room for a number of years. As we’ve learnt from the MDGs, to improve performance in such indicators it is enough to force people into poor buildings they call schools (threatening withdrawal of other type of support, e.g. CCTs, for non-attendance); increase class size; or to reduce the teachers/students ratio. Therefore, improvement in these indicators may not be connected with changes in the poverty of people, not even in terms of accumulation of human capital. In fact, none of these indicators measure what people living in poverty across the globe felt they needed. Recent participatory research by CAFOD revealed how the quality of education, adequate facilities and staff for students with disabilities were seen as priorities for people. Research participants thought they were wasting their children’s time by placing them in poorly equipped and understaffed schools.

Finally, the post-2015 framework has to help us reach the remaining half of the poor, those who are most marginalised. However, a lot of research, including CAFOD’s recent participatory study has demonstrated that often disadvantage and discrimination related to disability, gender and age takes place within the household, which is not captured by the MPI, mostly built upon household survey data.

In conclusion, many governments have finally understood the need for  multiple indicators to analyse and respond to poverty with appropriate policy. But do we really need to create a global arbitrary aggregate index of so many different variables? Should we add up apples and oranges? As some governments have clearly said during the 2013 UN General Assembly, we need to involve citizens directly in order to understand their perspectives and experiences of poverty. This requires a complex set of new indicators. While providing potentially valuable information, the MPI 2.0 should not become the central post-2015 indicator.

Tags: , , ,

6 Responses to “Do we need the Multidimensional Poverty Index (MPI) for post-2015?”

  1. Kim Gabrielli Says:

    Reblogged this on Internasjonalen and commented:
    One of the concrete inputs of the Norwegian consultation on post-2015 (summary posted yesterday) was that there is a need of a multidimensional poverty index. CAFOD policy team gives us an excellent overview of the poverty debate on this blog

  2. New development goals: What is poverty? | Internasjonalen Says:

    […]… […]

  3. Do we need the Multidimensional Poverty Index? | #FNdebatt Says:

    […] Read the posted blog here […]

  4. Do we need the Multidimensional Poverty Index (MPI) for post-2015? | - what comes after the MDGs? Says:

    […] Article by Andrea Rigon, writing on the CAFOD policy team blog. […]

  5. Sabina Alkire, Director, OPHI Says:

    Dear Andrea,

    Thank you for your thoughtful comments on the MPI2015+; below are some reflections on your questions.

    Firstly, there will not be a single post-2015 indicator. There will be a dashboard of indicators. This is necessary and we fully support it! So why do we say that an MPI2015+ could add value as a headline indicator on that dashboard?

    Q: What’s the added value of aggregating into a single global indicator?

    Competition between multiple indicators can stall action. The Millennium Development Goals (MDGs) – which did not include indicators on sustainability – had over 60. What an MPI does is give prominence not to just one, but to a (limited) set of indicators. Looking at a report showing 12 different indicators (for example), you may get a fuzzy sense of whether poverty has gone up or down, but it’s rather rough. By combining people’s direct experiences in those indicators into an MPI2015+, we can see separate and important facets of poverty quite transparently – the incidence of poverty (what percentage of people are multidimensionally poor); the intensity of poverty (how many deprivations poor people face at once); and the composition of poverty (in which indicators and dimensions people are poor). An MPI2015+ would also allow us to see how different kinds of poverty interconnect, helping to break apart the silos of poverty eradication programmes and inform more cost-effective and better targeted policies.

    Q. How can an MPI help eradicate poverty?

    Such a rich seam of poverty data provides policymakers with many tools. An MPI2015+ can be disaggregated to show marginalized regions and social groups, helping to better target the poorest and most vulnerable. It would also reflect changes in indicators directly, making it a time-sensitive and effective monitoring tool. For example, an increase in people with access to clean water will show up in an index as soon as new data are collected.

    It’s not just OPHI that believes an MPI2015+ would add value. In September 2013, 20+ countries and institutions in the global Multidimensional Poverty Peer Network (MPPN) proposed that the UN adopt a new headline indicator of multidimensional poverty. This should be constructed from participatory inputs of those living in poverty regarding the dimensions of poverty they consider most important. To implement an MPI2015+ requires appropriate data; hopefully more frequent and better quality data, that reflects, for example, the quality of education, not merely school attendance.

    Q. Wouldn’t an MPI2015+ use arbitrary cut-offs and weights?

    All poverty measures use cut-offs and weights. Every MDG has a ‘cut-off’; for example, with the goal on primary education, the ‘cut-off’ is the completion of primary school. The MPI2015+ would use the cut-offs that are already being discussed and will be implemented anyway.

    Weights are also in use. For example, you might say that in some countries there is progress in 22 indicators and in others, in only 13; by counting them, you are effectively ‘equally weighting’ each indicator. Similarly, when governments allocate resources, they are ‘weighting’ goals against each other – but how?

    The weights and cut-offs of an MPI are transparent. They can be set through participatory and expert consultation, and because they are transparent, they can be publicly debated and changed. Indeed any final measure would be designed so that key policy points were robust to a range of weights. In contrast, with GDP or an income poverty measure, the weights (price adjustments) are largely hidden from view due to their complexity.

    Q. Would an MPI incentivise action on only the cheapest policies?

    If a measure is well constructed and policymakers go for the ‘easy wins’ first, people will at least see an improvement in that aspect of poverty. Also, the MPI has some interesting inbuilt policy incentives that go beyond the barely poor.

    You reduce an MPI score either by reducing the number of poor people or the average intensity of their poverty. So if the intensity of poverty goes down (i.e. policies target the very poorest people and reduce some of their deprivations but don’t eliminate their poverty entirely) the MPI score will still go down. This innovation of the Alkire Foster method of multidimensional measurement rewards policymakers for targeting the most vulnerable, not just moving the least poor people across a poverty line. This is very important for equitable outcomes, and it really happens in countries we have studied.

    Q. How does the MPI show disadvantage and discrimination within the household (e.g. by gender, age or disability) if it’s based on household survey data?

    Importantly, the MPI can already be broken down by ethnicity and region. For example, in India we saw that the poorest groups (scheduled tribes and Muslims) had slower reductions in multidimensional poverty compared to others, whereas in some regions like Potosi in Bolivia (the poorest region), the MPI reduced the fastest. If gendered data are available post-2015 (as OPHI very much hopes they will be), we could further break down the MPI2015+ by gender, age, occupational status, disability, and so on. In short, the measure is willing, but the data are weak! For this reason OPHI and the MPPN are calling for a data revolution to support the post-2015 development agenda.

    Some links that you and your readers may find useful:

    • Briefing on ‘Multidimensional Poverty and the Post-2015 MDGs’

    • Policy applications of the Alkire Foster method for multidimensional measurement

    • Multidimensional Poverty Peer Network – participants and activities

  6. The Chronic Poverty Report and post-2015: matching policy with people’s lives | Serpents and Doves: A development policy blog Says:

    […] Secondly, many people were frustrated by the report’s continued use of $1.25 a day as the metric through which poverty is measured. Although the authors acknowledged that poverty is multidimensional , they’ve stuck with the conventional measure. Likewise, I’d agree that we need to move beyond a simplistic measure to understand and address the complex realities of people’s lives, such as the multi-poverty indicator suggested by the OPHI, but one that involves citizens directly in order to understand their perspectives and experiences of pov…. […]

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: